Systems and methods for image feature extraction

This description relates to image feature extraction. In some examples, a system can include a keypoint detector and a feature list generator. The keypoint detector can be configured to upsample a keypoint score map to produce an upsampled keypoint score map. The keypoint score map can include feature scores indicative of a likelihood of at least one feature being present at keypoints in an image. The feature list generator can be configured to identify a subset of keypoints of the keypoints in the image using the feature scores of the up sampled keypoint score map, determine descriptors for the subset of keypoints based on a feature description map, and generate a keypoint descriptor map for the image based on the determined descriptors.

All-weather target detection method based on vision and millimeter wave fusion

An all-weather target detection method based on a vision and millimeter wave fusion includes: simultaneously acquiring continuous image data and point cloud data using two types of sensors of a vehicle-mounted camera and a millimeter wave radar; pre-processing the image data and point cloud data; fusing the pre-processed image data and point cloud data by using a pre-established fusion model, and outputting a fused feature map; and inputting the fused feature map into a YOLOv5 detection network for detection, and outputting a target detection result by non-maximum suppression. The method fully fuses millimeter wave radar echo intensity and distance information with the vehicle-mounted camera images. It analyzes different features of a millimeter wave radar point cloud and fuses the features with image information by using different feature extraction structures and ways, so that the advantages of the two types of sensor data complement each other.

AUTOMATED DEFECT CLASSIFICATION AND DETECTION

The present disclosure related to a computer-implemented training and prediction method for defect detection, classification and segmentation in image data. The training method comprises providing an ensemble of learning structures, each learning structure comprising a feature extractor module, a region proposal module, a detection module, and a segmentation module. Each learning structure is trained individually and validated. Learning structures whose validation prediction score exceeds a predetermined threshold score are selected and their predictions combined, using a parametrized ensemble voting structure.

METHOD, ELECTRONIC DEVICE, AND COMPUTER PROGRAM PRODUCT FOR DATA PROCESSING

Embodiments of the present disclosure provide a method, an electronic device, and a computer program product for data processing. The method may include receiving a data conversion strategy from a server. The method may further include determining, in response to receiving unstructured data from a field device, metadata of the unstructured data based on the received data conversion strategy to form a set of metadata. In addition, the method may include transmitting at least a part of the set of metadata to the server. According to embodiments of the present disclosure, edge computing can be performed on unstructured data, which not only enables timely processing of monitoring data, but also reduces computing load on the side of the server, thereby improving the user experience.

IMAGE FEATURE MATCHING METHOD AND RELATED APPARATUS, DEVICE AND STORAGE MEDIUM

In an image feature matching method, at least two images to be matched are acquired; a feature representation of each image to be matched is obtained by performing feature extraction on the image to be matched, wherein the feature representation comprises a plurality of first local features; transforming the first local features into first transformation features having a global receptive field of the images to be matched; and a first matching result of the at least two images to be matched is obtained by matching first transformation features in the at least two images to be matched.

IMAGING PROCESSING METHOD AND APPARATUS, ELECTRONIC DEVICE, AND STORAGE MEDIUM

The present disclosure relates to an image processing method and apparatus, an electronic device and a storage medium. The method includes performing feature extraction on an image to be processed to obtain a first feature map of the image to be processed and performing weight prediction on the first feature map to obtain a weight feature map of the first feature map. The weight feature map includes weight values of feature points in the first feature map. The method further includes performing feature value adjustment on the feature points in the first feature map based on the weight feature map to obtain a second feature map and determining a processing result of the image to be processed according to the second feature map. Embodiments of the present disclosure may improve the image processing accuracy.

TRANSMISSION LINE DEFECT IDENTIFICATION METHOD BASED ON SALIENCY MAP AND SEMANTIC-EMBEDDED FEATURE PYRAMID

The present disclosure provides a transmission line defect identification method based on a saliency map and a semantic-embedded feature pyramid, including the following steps: step 1: cleaning and classifying a dataset; step 2: generating a super-resolution image for a small target of a transmission line by using an Electric Line-Enhanced Super-Resolution Generative Adversarial Network (EL-ESRGAN) model; step 3: performing image saliency detection on the dataset by constructing a U2-Net; step 4: performing data augmentation on the dataset by using GridMask and random cutout algorithms based on a saliency map, and generating a classified dataset; and step 5: performing image classification on a normal set and a defect set by using a ResNet34 classification algorithm and a deep semantic embedding (DSE)-based feature pyramid classification network.

GLOBAL CONTEXT VISION TRANSFORMER

Vision transformers are deep learning models that employ a self-attention mechanism to obtain feature representations for an input image. To date, the configuration of vision transformers has limited the self-attention computation to a local window of the input image, such that short-range dependencies are modeled in the output. The present disclosure provides a vision transformer that captures global context, and that is therefore able to model long-range dependencies in its output.

Feature compression and localization for autonomous devices

Systems, methods, tangible non-transitory computer-readable media, and devices associated with object localization and generation of compressed feature representations are provided. For example, a computing system can access source data and target data. The source data can include a source representation of an environment including a source object. The target data can include a compressed target feature representation of the environment. The compressed target feature representation can be based on compression of a target feature representation of the environment produced by machine-learned models. A source feature representation can be generated based on the source representation and the machine-learned models. The machine-learned models can include machine-learned feature extraction models or machine-learned attention models. A localized state of the source object with respect to the environment can be determined based on the source feature representation and the compressed target feature ...

SYSTEMS AND METHODS FOR AI-ASSISTED SURGERY

Various embodiments of the invention provide systems and methods to assist or guide an arthroscopic surgery or other surgical procedure e.g., surgery of the shoulder, knee or hip. The method comprises steps of receiving an image from an interventional imaging device, identifying a feature in the image using an image recognition algorithm, overlaying the features on a video feed on a display device and making recommendations or suggestions to an operator based on the identified feature in the image.

Machine learning-based point cloud alignment classification

A method, system, and computer program product, all suitable for machine learning based point cloud alignment classification is disclosed. The method comprises the steps of obtaining at least two light detection and ranging (LiDAR) point clouds 602, processing the at least two LiDAR point clouds using at least one classifier network 604, obtaining at least one output dataset from the at least one classifier network 606, determining that the at least two LiDAR point clouds are misaligned based on the at least one output dataset 608, and performing a first action based on determining that the at least two LiDAR point clouds are misaligned 610. The at least one classifier network may comprise at least one of a pillar-based network (see figure 5b) or a kernel point convolution-based network (see figure 5c). The first action may comprise labelling the at least two LiDAR point clouds as misaligned, and/or updating a locality of a map based on labelling the at least two LiDAR point clouds as misaligned ...

METHOD AND APPARATUS WITH IMAGE ANALYSIS

A processor-implemented method with image analysis includes: receiving a test image; generating a plurality of augmented images by augmenting the test image; determining classification prediction values for the augmented images using a classifier; determining a detection score based on the classification prediction values; and determining whether the test image corresponds to anomaly data based on the detection score and a threshold.

META-OPTIC ACCELERATORS FOR OBJECT CLASSIFIERS

A system for identifying objects in images is provided. The system may include an optical front end and a digital back end. The optical front end includes a metalens that duplicates a received image into multiple images, and a metasurface that receives the duplicate images and outputs a feature map based on the received images. The feature map may be equivalent to the computationally expensive convolution operations previously performed by a neural network. The feature map is provided to the digital back end, which uses a neural network to classify the object. Because the feature map included the convolution operations, the digital back end can classify the object more quickly and using fewer computing resources than previous systems.

PROCESSING APPARATUS, INFORMATION PROCESSING METHOD AND RECORDING MEDIUM

An information processing apparatus (2) includes: an extracting unit (212) that extracts a key point (KP) of a target object as a target key point (KP1) from an input image (IMG1); and a calculating unit (213) that calculates, as a score (SC) of the input image from which the target key point is extracted, an index value related to a reliability of the target key point based on the target key point.

IMAGE PROCESSING METHOD, DEVICE, ELECTRONIC APPARATUS AND STORAGE MEDIUM

The present disclosure relates to an image processing method and device, an electronic apparatus and a storage medium, and the image processing method includes: acquiring an input image; detecting a target area in the input image; and processing the target area, wherein the processing of the target area includes: obtaining a feature map of the target area, rearranging feature blocks in the feature map in a feature space, and obtaining an output image after the target area is processed based on the rearranged feature blocks and the feature map.

METHOD AND APPARATUS FOR FEATURE TRANSFORMING AND PROCESSING BASED ON ARTIFICIAL NEURAL NETWORK

Disclosed herein are a method and apparatus for processing feature information based on an artificial neural network. According to an embodiment of the present disclosure, the apparatus for processing feature information based on an artificial neural network may include a memory for storing data and a processor for controlling the memory, and the processor may further be configured to extract a graph, which includes vertices, based on a feature map of an image, to extract a feature vector corresponding to the vertices and to process the graph and the feature vector based on an artificial neural network, and the graph may include positions of the vertices and information on a connection relationship between the vertices.

Method and apparatus for asynchronous data fusion, storage medium and electronic device

A method and an apparatus for asynchronous data fusion, a storage medium and an electronic device are provided. The method includes: obtaining current frame LiDAR data, and determining current frame LiDAR three-dimensional embeddings; determining a previous frame fused hidden state, and performing a temporal fusion process based on the previous frame fused hidden state and the current frame LiDAR three-dimensional embeddings to generate a current frame temporary hidden state and a current frame output result; and obtaining current frame camera data, determining current frame camera three-dimensional embeddings, and generating a current frame fused hidden state based on the current frame camera three-dimensional embeddings and the current frame temporary hidden state. Asynchronous fusion is performed on the current frame LiDAR data and previous frame camera data, which leads to a low processing latency.

IMAGE RECOGNITION METHOD, IMAGE RECOGNITION APPARATUS, AND NON-TRANSITORY COMPUTER READABLE RECORDING MEDIUM STORING AN IMAGE RECOGNITION PROGRAM

An image recognition method includes a feature amount extracting step of generating, from an input image, a base feature map group including a plurality of base feature maps; an inferring step of deriving a plurality of inference results using each of a plurality of machine-learned inference devices for a plurality of inference inputs based on the base feature map group; and an integrating step of integrating the plurality of inference results by a specific manner to derive a final inference result, where each of the plurality of inference inputs has some or all base feature maps of the plurality of base feature maps, and each of the plurality of inference inputs has the some or all base feature maps that are different in part or whole from the some or all base feature maps of another inference input in the plurality of inference inputs.

Method, device, and computer program product for error evaluation

Embodiments of the present disclosure provide a method, device, and computer program product for error evaluation. A method for error evaluation comprises in accordance with a determination that an error occurs in a data protection system, obtaining context information related to an operation of the data protection system; determining, based on the context information and using a trained deep learning model, a type of the error in the data protection system from a plurality of predetermined types, the deep learning model being trained based on training context information and a label on a ground-truth type of an error associated with the training context information; and providing the determined type of the error in the data protection system. In this way, it is possible to achieve automatic classification of errors in the data protection system, thereby improving the efficiency in error classification and saving the operation costs. Therefore, more rapid and more accurate measures can ...

ESTIMATION MODEL FOR INTERACTION DETECTION BY A DEVICE

A method and device are disclosed for estimating an interaction with the device. The method includes configuring a first token and a second token of an estimation model according to first features of a 3D object, applying a first weight to the first token to produce a first-weighted input token and applying a second weight that is different from the first weight to the second token to produce a second-weighted input token, and generating, by a first encoder layer of an estimation-model encoder of the estimation model, an output token based on the first-weighted input token and the second-weighted input token. The method may include receiving, at a 2D feature extraction model, the first features from a backbone, extracting, by the 2D feature extraction model, second features including 2D features, and receiving, at the estimation-model encoder, data generated based on the 2D features.

METHOD AND APPARATUS FOR DETECTING CARGO IN CONTAINER IMAGE USING CONTAINER WALL BACKGROUND REMOVAL

Provided are a method and apparatus for detecting cargo in a container image using container wall background removal. The method of detecting cargo in a container image according to an aspect of the present invention includes: receiving a backscatter X-ray target container image captured on a target container; acquiring a backscatter X-ray empty target container wall image of an empty container that corresponds to the target container and a capturing condition of the backscatter X-ray target container image; generating a difference image in which a wall background of the target container is removed based on the backscatter X-ray target container image and the backscatter X-ray empty target container wall image; and detecting cargo included in the target container based on the difference image.

Methods and systems for determining the authenticity of an identity document

A method for determining the authenticity of an identity document is provided that includes capturing, by an electronic device, image data of an identity document, determining a class of the identity document, and extracting, using multi-resolution convolution and octave convolution techniques, first and second frequency components from the captured image data. The first and second frequency components correspond to different spatial frequency ranges. Moreover, the method includes determining whether the first and second frequency components satisfy matching criteria with data in corresponding frequency maps. The frequency maps are created from verified documents belonging to the determined class of document. In response to determining at least one of the first and second frequency components satisfies the matching criteria, determining the identity document is genuine.

Machine learning-based point cloud alignment classification

A method, system, and computer program product, all suitable for machine learning based point cloud alignment classification is disclosed. The method comprises the steps of obtaining at least two light detection and ranging (LiDAR) point clouds 602, processing the at least two LiDAR point clouds using at least one classifier network 604, obtaining at least one output dataset from the at least one classifier network 606, determining that the at least two LiDAR point clouds are misaligned based on the at least one output dataset 608, and performing a first action based on determining that the at least two LiDAR point clouds are misaligned 610. The at least one classifier network may comprise at least one of a pillar-based network (see figure 5b) or a kernel point convolution-based network (see figure 5c). The first action may comprise labelling the at least two LiDAR point clouds as misaligned, and/or updating a locality of a map based on labelling the at least two LiDAR point clouds as misaligned ...

Training Recognition Device

Provided is a training recognition device that implements training of a DNN for article recognition that does not require manual annotation for an image for training and can reduce power consumption, time, and hardware amount required for training. The training recognition device includes: an image conversion unit that inputs a simulation image and an actual site image into a generative adversarial network and converts the simulation image into an artificial site image; a pre-trained feature extraction unit that inputs the simulation image to a trained deep neural network trained using the simulation image and annotation data for the simulation image and outputs a feature point of the simulation image at time of re-training; a re-training feature extraction unit that inputs the artificial site image to a deep neural network for re-training, re-trains a difference between the simulation image and the artificial site image, and outputs a feature point of the artificial site image; an error ...

DEVICE AND METHOD FOR SYNTHESIZING IMAGE CAPABLE OF IMPROVING IMAGE QUALITY

An image synthesis device according to a disclosed embodiment has one or more processors and a memory which stores one or more programs executed by the one or more processors. The image synthesis device includes a first artificial neural network provided to learn each of a first task of using a damaged image as an input to output a restored image and a second task of using an original image as an input to output a reconstructed image, and a second artificial neural network connected to an output layer of the first artificial neural network, and trained to use the reconstructed image output from the first artificial neural network as an input and improve the image quality of the reconstructed image.

Systems and methods for processing electronic images of slides for a digital pathology workflow

A computer-implemented method of using a machine learning model to categorize a sample in digital pathology may include receiving one or more cases, each associated with digital images of a pathology specimen; identifying, using the machine learning model, a case as ready to view; receiving a selection of the case, the case comprising a plurality of parts; determining, using the machine learning model, whether the plurality of parts are suspicious or non-suspicious; receiving a selection of a part of the plurality of parts; determining whether a plurality of slides associated with the part are suspicious or non-suspicious; determining, using the machine learning model, a collection of suspicious slides, of the plurality of slides, the machine learning model having been trained by processing a plurality of training images; and annotating the collection of suspicious slides and/or generating a report based on the collection of suspicious slides.

METHOD AND APPARATUS WITH OBJECT RECOGNITION

A method and apparatus for object recognition are provided. A processor-implemented method includes extracting feature maps including local feature representations from an input image, generating a global feature representation corresponding to the input image by fusing the local feature representations, and performing a recognition task on the input image based on the local feature representations and the global feature representation.

MULTI-TASK OBJECT DETECTION METHOD, ELECTRONIC DEVICE, MEDIUM, AND VEHICLE

The disclosure provides a multi-task object detection method, an electronic device, a medium, and a vehicle, to solve the technical problem of low detection accuracy or poor detection effect of an existing multi-task detection method. For this purpose, the multi-task object detection method of the disclosure includes: obtaining images captured by a vehicle-mounted sensor; inputting the images into a multi-scale feature extraction network to extract multi-scale features; inputting the multi-scale features into a multi-scale feature fusion network to obtain fused features, where the multi-scale feature fusion network includes multiple optimal fusion paths, and each optimal fusion path corresponds to one of multiple tasks; and inputting, into a corresponding detection head, the fused features output from each optimal fusion path, to obtain a detection result, where each detection head is capable of detecting one of the multiple tasks. In this way, the accuracy of multi-task object detection ...

Obtaining patterns for surfaces of objects

Techniques to train a neural network using transformations

Neural network training images are randomly transformed (504). The training images, which may be generated from a first source (501), may be partial images cropped from a set of 3D images. The trained NNs may identify objects in images generated by a second, different, source (508). The transformation may comprise applying processing function(s) to the NN, e.g. associated with a probability and a magnitude. The NN training image may comprise image data and an annotation, both of which may be transformed. The transformation may comprise changing sharpness, blurriness, noise level, brightness, contrast, intensity perturbation, rotation, scaling and deformation. These transformations may be grouped into image quality, image appearance and spatial configuration. Also disclosed are neural networks being trained using stacked transformed images where the trained neural network is provided for processing images from an unseen domain distinct from a source domain, wherein stacked transformed images ...

SALIENCY-BASED INPUT RESAMPLING FOR EFFICIENT OBJECT DETECTION

A processor-implemented method of video processing using includes receiving, via an artificial neural network (ANN), a video including a first frame and a second frame. A saliency map is generated based on the first frame of the video. The second frame of the video is sampled based on the saliency map. A first portion of the second frame is sampled at a first resolution and a second portion of the second frame is sampled at a second resolution. The first resolution is different than the second resolution. A resampled second frame is generated based on the sampling of the second frame. The resampled second frame is processed to determine an inference associated with the video.

METHOD AND SYSTEM FOR IMAGE CLASSIFICATION

There is provided a method of image classification. The method includes: providing a set of category mapping discriminators, each corresponding to a respective category, wherein each category mapping discriminator of the set of category mapping discriminators is configured for discriminating features relating to input images that belong to the respective category of the category mapping discriminator; extracting a plurality of features from an input image using a machine learning model; determining, for each of the set of category mapping discriminators, an output value based on the plurality of extracted features using the category mapping discriminator; and determining a classification of the input image based on the output values of the set of category mapping discriminators.

SEQUENCE RECOGNITION METHOD AND APPARATUS, ELECTRONIC DEVICE, AND STORAGE MEDIUM

A sequence recognition method is implemented by using a sequence recognition network. The sequence recognition network at least includes an encoder network and a decoder network. The method includes: acquiring a to-be-processed image, the to-be-processed image including a to-be-recognized object sequence; encoding the to-be-processed image by using the encoder network to obtain a first feature sequence; decoding the first feature sequence by using the decoder network to obtain a second feature sequence; and obtaining a sequence recognition result of the object sequence based on the second feature sequence, where the sequence recognition network is obtained by respectively supervising the encoder network and the decoder network.

Spatially varying reduction of haze in images

Methods, systems, devices, and tangible non-transitory computer readable media for haze reduction are provided. The disclosed technology can include generating feature vectors based on an input image including points. The feature vectors can correspond to feature windows associated with features of different portions of the points. Based on the feature vectors and a machine-learned model, a haze thickness map can be generated. The haze thickness map can be associated with an estimate of haze thickness at each of the points. Further, the machine-learned model can estimate haze thickness associated with the features. A refined haze thickness map can be generated based on the haze thickness map and a guided filter. A dehazed image can be generated based on application of the refined haze thickness map to the input image. Furthermore, a color corrected dehazed image can be generated based on performance of color correction operations on the dehazed image.

Systems, devices, and methods for projecting digital content including hair color changes onto a user's head, face, or body

In an embodiment, a virtual hair coloration system includes: a projector 22 configured to project digital content including a makeup application tutorial onto the user's hair; and a dynamic mapping unit 24; 30 operably coupled to the projector, wherein the dynamic mapping unit is configured to establish a dynamic correspondence between pixels of the projector 22 and features of the user's hair.

VIDEO TRANSFORMER FOR DEEPFAKE DETECTION WITH INCREMENTAL LEARNING

A method, apparatus, and system for detecting DeepFake videos, includes an input device for inputting a potential DeepFake video, the input device inputs a sequence of video frames of the video, and processing circuitry. The processing circuitry detects faces frame by frame in the video to obtain consecutive face images, creates UV texture maps from the face images, inputs both face images and corresponding UV texture maps, extracts image feature maps, by a convolution neural network (CNN) backbone, from the input face images and corresponding UV texture maps and forms an input data structure, receives the input data structure, by a video transformer model that includes multiple encoders, and computes, by the video transformer model, a classification of the video as being Real or Fake. A display device plays back the potential DeepFake video and an indication that the video is Real or Fake.

PRODUCT POSITIONING METHOD

PATIENT RESPONSE-BASED BIOMARKER TOPOLOGY QUANTIFICATION AND ASSESSMENT FOR MULTIPLE TISSUE TYPES

TRANSLATION OF TEXT DEPICTED IN IMAGES

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, that translate text depicted in images from a source language into a target language. Methods can include obtaining a first image that depicts first text written in a source language. The first image is input into an image translation model, which includes a feature extractor and a decoder. The feature extractor accepts the first image as input and in response, generates a first set of image features that are a description of a portion of the first image in which the text is depicted is obtained. The first set of image features are input into a decoder. In response to the input first set of image features, the decoder outputs a second text that is a predicted translation of text in the source language that is represented by the first set of image features.

IMAGING APPARATUS, IMAGING CONTROL METHOD, AND PROGRAM

An imaging apparatus according to the present technology includes a mode control unit. The mode control unit shifts, when a motion is detected on a motion detection mode to detect the motion on the basis of image information, the mode to a feature detection mode to detect features on the basis of image information having a higher resolution than a resolution of the image information that is used for the motion detection, and shifts, when a specific feature is detected on the feature detection mode, the mode to an imaging mode to acquire image information having a higher resolution than the resolution of the image information that is used for the feature detection.

Method for generating web code for UI based on a generative adversarial network and a convolutional neural network

Provided is a method for generating web codes for a user interface (UI) based on a generative adversarial network (GAN) and a convolutional neural network (CNN). The method includes steps described below. A mapping relationship between display effects of a HyperText Markup Language (HTML) element and source codes of the HTML element is constructed. A location of an HTML element in an image I is recognized. Complete HTML codes of the image I are generated. The similarity between manually-written HTML codes and the generated complete HTML codes and the similarity between the image I and an image I1generated by the generated complete HTML codes are obtained. After training, an image-to-HTML-code generation model M is obtained. A to-be-processed UI image is input into the model M so as to obtain corresponding HTML codes. According to the method of the present disclosure, an image-to-HTML-code generation model M can be obtained.

Cell classification algorithm

A method of detecting a protein (e.g. receptor) on cells by obtaining spatial coordinates of the detected protein. The method comprises detecting boundaries of the cells. A data vector is constructed based on the obtained spatial coordinates and the detected boundaries. The method can use dSTORM or fPALM. A spatial distribution can be evaluated based on the obtained spatial coordinates. The spatial coordinates can be partitioned into one or more clusters at predetermined length scales by performing a spatial distribution analysis algorithm. The boundaries can be obtained by use of an optical image of the cells. A segmentation algorithm can be performed on the optical image of the cells. A border obtained by the segmentation algorithm can be extended by a predetermined distance. Colocalization analysis can be performed on an overlapping area between two cells. Principal component analysis (PCA) can be performed on the data. The cells can be classified based on a reference cell. A partitioning ...

Feature fusion method and apparatus for image processing, electronic device and storage medium

The present disclosure provides an image processing method. An image to be classified is input into a feature extraction model to generate N dimensional features. Dimension fusion is performed on M features of the N dimensional features to obtain M dimension fusion features. The image to be classified is processed based on M dimension fusion features and remaining features of the N dimensional features other than the M features.

Machine learning-based root cause analysis of process cycle images

The technology disclosed relates to classification of process cycle images to predict success or failure of process cycles. The technology disclosed includes capturing and processing images of sections arranged on an image generating chip in genotyping process. Image description features of production cycle images are created and given as input to classifiers. A trained classifier separates successful production images from unsuccessful or failed production images. The failed production images are further classified by a trained root cause classifier into various categories of failure.

SYSTEMS AND METHODS FOR HYPERSPECTRAL IMAGING AND ARTIFICIAL INTELLIGENCE ASSISTED AUTOMATED RECOGNITION OF DRUGS

This disclosure relates to a system and a method for automated recognition of drugs. This disclosure also relates to a system for automated recognition of drugs comprising a hyper-spectral imaging system. This disclosure also relates to a hyper-spectral imaging system configured to automatically recognize drugs by using a neural network. This disclosure relates to training the neural network to identify a drug type (e.g., the name of the drug) based on an image (e.g., normal visible image and/or hyperspectral image) of the drug.

PERSON MOVEMENT TYPE DETERMINATION METHOD, PERSON MOVEMENT TYPE DETERMINATION DEVICE, AND STORAGE MEDIUM

A person movement type determination method includes: acquiring time-series data of a position or a speed of a predetermined feature portion which is a part of a body of a target person; extracting a feature value of a predetermined frequency component or a predetermined frequency band based on oscillation of the predetermined feature portion indicating a movement type to be determined from the time-series data; and determining whether a movement type of the target person is that which is to be determined based on the feature value.

TEACHER DATA COLLECTING METHOD AND COLLECTING DEVICE

A teacher data collecting method in a defect classification model for classifying a defect by using, as teacher data, a few pieces of expert data and many pieces of non-expert data, includes: encoding, into one dimension, a latent variable of a variational auto encoder that has been caused to perform learning the expert data; inputting the non-expert data into the variational auto encoder and encoding a latent variable into one dimension; calculating maximum values and minimum values of the latent variable in one dimension of the expert data and the non-expert data; and determining whether to complete collection of the non-expert data, based on a ratio of a difference between the maximum value and the minimum value of the latent variable in one dimension of the non-expert data to a difference between the maximum value and the minimum value of the latent variable in one dimension of the expert data.

AUTOMATICALLY GENERATING BEST DIGITAL IMAGES OF A PERSON IN A PHYSICAL ENVIRONMENT

The disclosed technology provides for generating best images of a person in a retail environment. A method may include receiving, by an edge computing device from a camera, a continuous stream of image data of the retail environment, detecting, using object detection techniques, a person in the image data, the image data including a group of images that are part of a time series, generating bounding boxes for each of the group of images around the person based on detecting the person as they move in the images, identifying, based on applying a features model to each bounding box, at least one feature of the group of images depicting the person, selecting a subset of the bounding boxes having at least one feature that satisfies best images criteria, the subset having best images of the person, and returning the best images of the person.

OBJECT RE-IDENTIFICATION USING POSE PART BASED MODELS

An example apparatus for re-identifying objects includes an image receiver to receive a first image and a second image of an object with an identity. The apparatus also includes a fused model generator to fuse a global representation of the object with local representations of pose parts of the object to generate a fused representation of the object based on the first image. The apparatus further includes an object re-identifier to re-identify the object with the identity in the second image based on the fused representation.

LEARNING DEVICE, OBJECT DETECTION DEVICE, LEARNING METHOD, AND RECORDING MEDIUM

A learning device makes an object detection device learn how to detect an object from an input image. A feature extraction unit performs feature extraction from input images including real images and pseudo images to generate feature maps, and the object detection unit detects objects included in the input images based on the feature maps. The domain identification unit identifies the domains forming the input images and generates domain identifiability information. Then, the feature extraction unit and the object detection unit learn common features that do not depend on the difference in domains, based on the domain identifiability information.

COMPUTER VISION-BASED SURGICAL WORKFLOW RECOGNITION SYSTEM USING NATURAL LANGUAGE PROCESSING TECHNIQUES

Systems, methods, and instrumentalities are disclosed for computer vision-based surgical workflow recognition using natural language processing (NLP) techniques. Surgical video of surgical procedures may be processed and analyzed, for example, to achieve workflow recognition. Surgical phases may be determined based on the surgical video and segmented to generate an annotated video representation. The annotated video representation of the surgical video may provide information associated with the surgical procedure. For example, the annotated video representation may provide information on surgical phases, surgical events, surgical tool usage, and/or the like.

METHODS OF PERFORMING REAL-TIME OBJECT DETECTION USING OBJECT REAL-TIME DETECTION MODEL, PERFORMANCE OPTIMIZATION METHODS OF OBJECT REAL-TIME DETECTION MODEL, ELECTRONIC DEVICES AND COMPUTER READABLE STORAGE MEDIA

The present disclosure relates to a method of performing real-time object detection using an object real-time detection model and a performance optimization method of object real-time detection model. According to an embodiment, the method of performing real-time object detection using an object real-time detection model includes: obtaining an identification image of a preset size by pre-processing an input image; obtaining object central point data and object size data by processing the identification image using the object real-time detection model; and obtaining an object detection result by determining an object region in the input image according to the object central point data and the object size data.

Method and System for Multi-Scale Vision Transformer Architecture

A computer-implemented method for processing images in deep neural networks by: breaking an input sample into a plurality of non-overlapping patches; converting said patches into a plurality of patch-tokens; processing said patch-tokens in at least one transformer block comprising a multi-head self-attention block; providing a multi-scale feature module block in the at least one transformer block; using said multi-scale feature module block for extracting features corresponding to a plurality of scales by applying a plurality of kernels having different window sizes; concatenating said features in the multi-scale feature module block; providing a plurality of hierarchically arranged convolution layers in the multi-scale feature module block; and processing said features in said hierarchically arranged convolution layers for generating at least three multiscale tokens containing multiscale information.

BUILDING ACCURATE TRAINING IMAGES FOR AUTOMATIC SEISMIC INTERPRETATION

Method and apparatus for video recognition

METHOD AND APPARATUS FOR DEEP LEARNING-BASED REAL-TIME ON-DEVICE AUTHENTICATION

Disclosed are a method and apparatus for real-time on-device authentication based on deep learning. A deep learning-based authentication method includes detecting a location of a region of interest (ROI) occupied by a face portion an input image by using a detection model, extracting a feature map from the input image by using a feature extractor of the detection model, extracting a fixed length feature for the face portion using the feature map and ROI pooling for the detected location of the ROI, and classifying a face included in the input image based on the fixed length feature.

DEVICE AND METHOD FOR TRAINING A NEURAL NETWORK FOR IMAGE ANALYSIS

A computer-implemented method for training a neural network. The training includes: determining a first feature map by the neural network based on a first transformed image, the first transformed image being determined based on a first transformation of a training image; determining a second feature map by the neural network based on a second transformed image, the second transformed image being determined based on a second transformation of the training image; determining a first loss value characterizing a metric between a first feature vector of the first feature map and a weighted sum of second feature vectors of the second feature map, weights of the weighted sum being determined according to overlaps of a part of the training image characterized by the first feature vector with respect to parts of the training image characterized by the respective second feature vectors; training the neural network based on the first loss value.

METHOD AND DEVICE FOR TRAINING A STYLE ENCODER OF A NEURAL NETWORK AND METHOD FOR GENERATING A DRIVING STYLE REPRESENTATION REPRESENTING A DRIVING STYLE OF A DRIVER

A method for training a style encoder of a neural network. Sensory input variables, which represent a movement of a system and surroundings of the system, are compressed to an abstract driving situation representation in at least one portion of a latent space of the neural network, using a trained situation encoder of the neural network. The sensory input variables are compressed to a driving style representation in at least one portion of the latent space, using the untrained style encoder. The driving style representation and the driving situation representation are decompressed from the latent space to output variables, using a style decoder of the neural network. A structure of the style encoder is changed to train the style encoder until the output variables of the style decoder represent the movement.

INFORMATION GENERATING METHOD AND APPARATUS, DEVICE, STORAGE MEDIUM, AND PROGRAM PRODUCT

An information generating method is performed by a computer device. The method includes: obtaining a target image; extracting a semantic feature set and a visual feature set of the target image; performing attention fusion on semantic features and visual features of the target image at n time steps to obtain caption words of the target image at the n time steps by processing the semantic feature set and the visual feature set of the target image through an attention fusion network in an information generating model; and generating image caption information of the target image based on the caption words of the target image at the n time steps. Through the foregoing method, an advantage of the visual feature in generating visual vocabulary and an advantage of the semantic feature in generating a non-visual feature are combined, thereby improving the image caption's accuracy.

System and method for property typicality determination

The method for property typicality determination can include: determining a property, determining attribute values for the property, determining a reference population for the property, determining reference population attribute values, determining a typicality metric for the property, and optionally determining an influential attribute.

Method and apparatus for video recognition

A method of performing video recognition using a transformer machine learning model, comprising: receiving a video (Input video); defining a temporal window comprising at least three frames of the video (central column, Fig.1D); and using the transformer-based ML model to perform a single spatial attention over the frames in the temporal window to identify a feature (centre hatched square, Fig.1D). Defining the temporal window may comprise dividing each frame into patches and selecting patches located at identical positions within each of the at least three frames. A classification token may be attached to the patch sequence. The spatial attention may be performed by inputting the patch sequence and classification token into the transformer deep learning model. The method may identify objects, actions or gestures in video and provide feedback to a user on the identified feature (Fig.7). The method may receive a user query and identify a sequence matching the query in stored video (Fig.6 ...

DISMANTLING PROCEDURE SELECTING APPARATUS, DISMANTLING PROCEDURE SELECTING METHOD, AND DISMANTLING APPARATUS

A dismantling procedure selecting apparatus includes a dismantling information storage unit that stores a plurality of pieces of dismantling information respectively including a plurality of predetermined dismantling procedures for a plurality of dismantled objects each used as a reference for dismantling an object to be dismantled, a whole detector that captures a whole image of the object to be dismantled, a detail detector that captures an image of at least a portion of the object to be dismantled, and a dismantling procedure deriving unit that extracts a first feature as a feature of the object to be dismantled based on data obtained by capturing an image by at least one of the whole detector and the detail detector, obtains a degree of matching between the first feature and each of a plurality of second features that are respectively features of a plurality of dismantled objects, and selects a dismantling procedure associated with one of a plurality of the dismantled objects having ...

MODEL GENERATING APPARATUS AND MODEL GENERATING METHOD

A model generating apparatus includes a measurement unit and a change unit. The measurement unit measures a size of an object appearing on an image included in data set. The change unit changes, based on a distribution of the size, a layer to be connected to a detection unit of a Convolutional Neural Network (CNN) for detecting an object appearing on the image, among layers for extracting feature included in the CNN.

FEW-SHOT DEFECT DETECTION METHOD BASED ON METRIC LEARNING

A few-shot defect detection method based on metric learning, including: (S1) performing data enhancement on a to-be-detected few-shot defect data set through a G2-Generative adversarial network (G2-GAN); (S2) extracting features of a defect data set similar to the to-be-detected few-shot defect data set based on an adaptive convolution kernel-based convolutional neural network (SKM-CNN) to generate a pre-training model; and (S3) transferring the pre-training model to a few-shot defect detection network (S2D2N) based on metric learning; and performing target feature extraction and metric learning in sequence to realize rapid identification and location of defects.

WIRELESS TRANSMITTER IDENTIFICATION IN VISUAL SCENES

Wireless transmitter identification in visual scenes is provided. This technology enables important wireless communications and sensing applications such as (i) fast beam/blockage prediction in fifth generation (5G)/sixth generation (6G) systems using camera data, (ii) identifying cars and people in a surveillance camera feed using joint visual and wireless data processing, and (iii) enabling efficient autonomous vehicle communication relying on both the camera and wireless data. This is done by developing multimodal machine learning based frameworks that use the sensory data obtained by visual and wireless sensors. More specifically, given some visual data, an algorithm needs to perform the following: (i) predict whether an object responsible for a received radio signal is present or not, (ii) if it is present, detect which object it is out of the candidate transmitters, and (iii) predict what type of signal the detected object is transmitting.

Target Tracking Method and Device, and Electronic Apparatus

The present disclosure provides a target tracking method, a target tracking device and an electronic apparatus. The target tracking method includes: inputting an iimage and an (i−1)image in a to-be-detected video stream into a target deep learning model, i being an integer greater than 1; detecting a target in the iimage to obtain a first target detection box, and tracking the target in the (i−1)image to obtain a tracking heatmap; and determining a target tracking result in accordance with the first target detection box, the tracking heatmap and the (i−1)image. 1. A target tracking method , comprising:{'sup': th', 'th, 'inputting an iimage and an (i−1)image in a to-be-detected video stream into a target deep learning model, i being an integer greater than 1;'}{'sup': th', 'th, 'detecting a target in the iimage to obtain a first target detection box, and tracking the target in the (i−1)image to obtain a tracking heatmap; and'}{'sup': 'th', 'determining a target tracking result in accordance with the first target detection box, the tracking heatmap and the (i−1)image.'}2. The target tracking method according to claim 1 , wherein determining the target tracking result comprises:determining a target tracking heatmap in the tracking heatmap in accordance with an index of an anchor box corresponding to the first target detection box and coordinates of a center of the first target detection box; and{'sup': 'th', 'determining coordinates of a center of the target in the first target detection box on the (i−1)image in accordance with coordinates of a point with a maximum value in the target tracking heatmap,'}{'sup': 'th', 'wherein the target tracking result comprises the coordinates of the center of the target in the first target detection box on the (i−1)image and the coordinates of the center of the first target detection box.'}3. The target tracking method according to claim 2 , wherein subsequent to determining the coordinates of the center of the target in the first ...

MULTIPLE OBJECT DETECTION METHOD AND APPARATUS

Disclosed are multiple object detection method and apparatus. The multiple object detection apparatus includes a feature map extraction unit for extracting a plurality of multi-scale feature maps based on an input image, and a feature map fusion unit for generating a multi-scale fusion feature map including context information by fusing adjacent multi-scale feature maps among the plurality of multi-scale feature maps generated by the feature map extraction unit.

LARGE-SCALE CROP PHENOLOGY EXTRACTION METHOD BASED ON SHAPE MODEL FITTING METHOD

Disclosed is a large-scale crop phenology extraction method based on a shape model fitting method. The method comprises: acquiring a multi-year vegetation index time sequence curve in a localized geographic region; performing smooth fitting on the vegetation index time sequence curve by using a dual logistic function fitting means; establishing shape models by using reference curves and reference points of agrometeorological stations; performing shape model fitting by means of transformation; and obtaining a phenological period extraction value of the localized geographic region by means of calculation using the optimal scaling parameter. According to the present invention, macroscopic features of the curve are used, such that the influence of localized fluctuation and noise of the curve can be reduced, and a better extraction precision is obtained; and each phenological period of a crop can be extracted at the same time.

IMAGE OPTIMIZATION METHOD AND SYSTEM BASED ON ARTIFICIAL INTELLIGENCE

The present application discloses an image optimization method and system based on artificial intelligence, comprising: performing image object recognition for a template image, and extracting and recording template information of the template image; identifying the number and categories of primary objects in a to-be-optimized image by using an image recognition algorithm of a neural network; performing matching in a template database for the to-be-optimized image according to a set matching condition and the recorded template information to obtain optimal template information; scaling the primary object in the to-be-optimized image according to layout information in the optimal template information; cropping and adjusting a background part of the to-be-optimized image according to resolution requirements of an output target image and a basic composition principle; combining an image of the processed primary object with a background image according to the layout information in the optimal ...

METHOD AND SYSTEM FOR AUTOMATED EVALUATION OF ANIMALS

Embodiments herein generally relate to a method and system for automated evaluation of animals. In at least one embodiment, the method comprises: accessing sensor data acquired of an animal; analyzing the sensor data to generate derivative sensor data; applying feature extraction to one or more of the sensor data and derivative sensor data to extract trait-specific feature data associated with the one or more target traits use for evaluating the animal; and generating one or more evaluation scores for the animal, for each of the one or more target traits, based on the extracted trait-specific feature data for these target traits.

SUPER RESOLUTION DEVICE AND METHOD

A super resolution device and method are provided. The device comprises an input interface and a sharing layer calculator group. An image and a scaling signal are received by the input interface, wherein the scaling signal is configured to indicate to perform a double scaling operation or a quadruple scaling operation to the image. When the scaling signal indicates to perform the quadruple scaling operation, the sharing layer calculator group performs a plurality of convolution operations based on a first number of input channels and a first number of output channels. When the scaling signal indicates to perform the double scaling operation, the sharing layer calculator group performs the convolution operations based on a second number of input channels and a second number of output channels.

NEXT GENERATION QUALITY INSPECTION

Methods and systems for inspecting a product, such as a wire harness, including product features for inspection. A camera of an inspection station may capture a product image. A machine learning (ML) model may detect one or more objects in the captured product image and provide, for each detected object, an identification of a class of the detected object and an identification of a region of the detected object in the captured product image. The class of the detected object may be either an acceptable product feature class or an unacceptable product feature class. The inspection station may display an enhanced product image that includes the captured product image to which the identification of the class of the detected object and the identification of the region of the detected object in the captured product image for each detected object have been added.

SPLIT ARCHITECTURE FOR ARTIFICIAL INTELLIGENCE-BASED BASE CALLER

Techniques for Producing Three-Dimensional Models from One or More Two-Dimensional Images

Described are techniques for producing a three-dimensional model of a scene from one or more two dimensional images. The techniques include receiving by a computing device one or more two dimensional digital images of a scene, the image including plural pixels, applying the received image data to scene generator/scene understanding engine that produces from the one or more digital images a metadata output that includes depth prediction data for at least some of the plural pixels in the two dimensional image and that produces metadata for a controlling a three-dimensional computer model engine, and outputting the metadata to a three-dimensional computer model engine to produce a three-dimensional digital computer model of the scene depicted in the two dimensional image.

IMAGE PROCESSING METHOD AND RELATED DEVICE

Methods and apparatuses for performing image processing using artificial intelligence are provided. In an embodiment, an image processing method includes acquiring a target image based on an editing operation on an original image, performing a filling processing operation on the target image to obtain a first filled image including a first target region, identifying a target patch based on the first filled image, calculating a similarity value between the target patch and at least one patch related to the first filled image using a first artificial intelligence (AI) model, determining a target residual patch corresponding to the target patch based on the similarity value and generating a processing result image based on the target residual patch.

Adaptable Automated Interpretation of Rapid Diagnostic Tests Using Self-Supervised Learning and Few-Shot Learning

A framework for a few-shot learning method is disclosed. In a first part, self-supervision and classification supervision are used to train a feature extractor. An example self-supervision method comprises running grayscale images through an edge filter, normalizing the filtered images, setting the normalized images to ground truth, generating feature-extracted images, using a decoder to reconstruct images from the feature-extracted images, determining a loss between the reconstructed images and the ground truth images, and using the loss to update parameters of the feature extractor. In a second part, a few-shot adaptation process is performed to adapt the model to a novel rapid test kit.

MACHINE LEARNING FRAMEWORK APPLIED IN A SEMI-SUPERVISED SETTING TO PERFORM INSTANCE TRACKING IN A SEQUENCE OF IMAGE FRAMES

A method and system are provided for tracking instances within a sequence of video frames. The method includes the steps of processing an image frame by a backbone network to generate a set of feature maps, processing the set of feature maps by one or more prediction heads, and analyzing the embedding features corresponding to a set of instances in two or more image frames of the sequence of video frames to establish a one-to-one correlation between instances in different image frames. The one or more prediction heads includes an embedding head configured to generate a set of embedding features corresponding to one or more instances of an object identified in the image frame. The method may also include training the one or more prediction heads using a set of annotated image frames and/or a plurality of sequences of unlabeled video frames.

SYSTEMS AND METHODS FOR REGISTRATION FEATURE INTEGRITY CHECKING

Systems and methods for registration feature integrity checking include a repositionable arm configured to support a repositionable device and a control unit. The control unit is configured to receive a feature set including one or more features extracted from one or more images of a repositionable structure obtained from an imaging device, determine an expected feature corresponding to an extracted feature in the feature set based on one or more models of the repositionable structure, determine an error between the extracted feature and the expected feature, determine whether to remove the extracted feature from the feature set based on the determined error, remove the extracted feature from the feature set in response to determining that the extracted feature should be removed from the feature set, and provide the feature set to a registration module. The repositionable structure includes the repositionable arm and/or the repositionable device.

INFORMATION PROCESSING DEVICE, INFORMATION PROCESSING METHOD, INFORMATION PROCESSING PROGRAM, AND INFORMATION PROCESSING SYSTEM

An information processing device () according to the present disclosure includes a first processing unit () and a second processing unit (). The first processing unit () includes a first feature amount extraction unit () and a second feature amount extraction unit (). The first feature amount extraction unit () executes, on data input from a sensor, feature amount extraction processing of extracting a feature amount of the data on the basis of a parameter learned through machine learning. The second feature amount extraction unit () executes, on reference data, feature amount extraction processing of extracting a feature amount of the reference data on the basis of the parameter. The second processing unit () includes a difference detection unit (). The difference detection unit () detects a difference between a first feature amount input from the first feature amount extraction unit () and a second feature amount input from the second feature amount extraction unit ().

Method and system for image classification

There is provided a method of image classification. The method includes: providing a set of category mapping discriminators, each corresponding to a respective category, wherein each category mapping discriminator of the set of category mapping discriminators is configured for discriminating features relating to input images that belong to the respective category of the category mapping discriminator; extracting a plurality of features from an input image using a machine learning model; determining, for each of the set of category mapping discriminators, an output value based on the plurality of extracted features using the category mapping discriminator; and determining a classification of the input image based on the output values of the set of category mapping discriminators.

A damage detection apparatus and method

METHOD AND ELECTRONIC DEVICE FOR DETECTING CANDID MOMENT IN IMAGE FRAME

Embodiments herein provide a method for detecting a candid moment in an image frame. The method includes: receiving, by an electronic device, image frames; determining, by the electronic device, a candid score of each image frame in the image frames using a Machine Learning (ML) model, wherein the candid score is a quantitative value of candidness present in the image frames; determining, by the electronic device, whether the candid score of the image frame in the image frames meets a threshold candid score; identifying, by the electronic device, that the candid moment is present in the image frame in response to determining that the candid score of the image frame meets the threshold candid score; and displaying, by the electronic device, the image frame comprising the candid moment. 1. A method for detecting a candid moment in an image frame , comprising:receiving, by an electronic device, a plurality of image frames;determining, by the electronic device, a candid score of each image frame in the plurality of image frames using at least one Machine Learning (ML) model, wherein the candid score is a quantitative value of candidness present in the plurality of image frames;determining, by the electronic device, whether the candid score of the at least one image frame in the plurality of image frames meets a threshold candid score;identifying, by the electronic device, that a candid moment is present in the at least one image frame in response to determining that the candid score of the at least one image frame meets the threshold candid score; anddisplaying, by the electronic device, the at least one image frame comprising the candid moment.2. The method as claimed in claim 1 , wherein determining claim 1 , by the electronic device claim 1 , the candid score of each image frame in the plurality of image frames using the at least one ML model claim 1 , comprises:identifying, by the electronic device, local features comprising at least one of a head pose, a gaze, a ...

Techniques for deriving and/or leveraging application-centric model metric

Techniques for quantifying accuracy of a prediction model that has been trained on a data set parameterized by multiple features are provided. The model performs in accordance with a theoretical performance manifold over an intractable input space in connection with the features. A determination is made as to which of the features are strongly correlated with performance of the model. Based on the features determined to be strongly correlated with performance of the model, parameterized sub-models are created such that, in aggregate, they approximate the intractable input space. Prototype exemplars are generated for each of the created sub-models, with the prototype exemplars for each created sub-model being objects to which the model can be applied to result in a match with the respective sub-model. The accuracy of the model is quantified using the generated prototype exemplars. A recommendation engine is provided for when there are particular areas of interest.

COMPUTER-READABLE RECORDING MEDIUM STORING INFORMATION PROCESSING PROGRAM, METHOD OF PROCESSING INFORMATION, AND INFORMATION PROCESSING APPARATUS

A non-transitory computer-readable recording medium stores an information processing program for causing a computer to execute a process including: extracting a first feature from an image; detecting, from the extracted first feature, a plurality of visual entities included in the image; generating a second feature in which the visual entities in at least one combination of the plurality of detected visual entities are combined, in first feature, with each other; generating, based on the first feature and the second feature, a first map that indicates relation of each visual entity; extracting a fourth feature based on the first map and a third feature obtained by converting the first feature; and estimating the relation from the fourth feature.

MULTISCALE POINT CLOUD CLASSIFICATION METHOD AND SYSTEM

The present disclosure discloses a multiscale point cloud classification method. The method includes the following steps: acquiring 3D unordered point cloud data; performing feature extraction and classification on the acquired point cloud data using a pre-trained parallel classification network to obtain an output result, wherein the parallel classification network includes a plurality of basic networks with the same structures; and fusing the output results of the parallel network using a pre-trained deep Q network to obtain a final result of point cloud classification. The present disclosure can improve the accuracy and robustness of point cloud classification.

FEATURE EXTRACTION METHOD, MODEL TRAINING METHOD, DETECTION METHOD OF FRUIT SPECTRUM

A feature extraction method of fruit spectrum includes taking a vector of each wavelength point in spectrum of samples as source data, and acquiring a sorting of all vectors by processing the source data by SPA; according to the sorting of the vectors, acquiring distribution points of each sample on a coordinate system; acquiring classification results of the samples by destructive analysis, and acquiring a number of first sample categories; acquiring a first Euclidean distance between the first sample categories; according to a sorting of the wavelength points, acquiring distribution points of each sample on the coordinate system; acquiring a number of second sample categories; acquiring a second Euclidean distance between the second sample categories; determining whether the first Euclidean distance is less than the second Euclidean distance; determine a (M+2)-th vector to be valid or invalid based on a comparison result.

ELECTRONIC DEVICE AND OPERATING METHOD THEREOF

An electronic device recognizing an object is provided. The electronic device includes a plurality of computing units, a memory, and a processor configured to control at least one of the plurality of computing units such that object information about objects obtained by recognizing the objects existing in a space by using a first recognition model, divide the space into a plurality of subset spaces, based on the object information, determine at least one recognition model, based on characteristic information of each of the subset spaces, assign the determined recognition model to one computing unit, based on characteristic information of each of a plurality of computing units and characteristic information of the determined recognition model, and control the plurality of computing units to perform object recognition by using the determined recognition model and the one computing unit in each of the subset spaces.

APPARATUS AND METHOD FOR DETECTING KEYPOINT BASED ON DEEP LEARNIING USING INFORMATION CHANGE ACROSS RECEPTIVE FIELDS

Disclosed herein are an apparatus and method for detecting a keypoint based on deep learning robust to scale changes based on information change across receptive fields. The apparatus for detecting a keypoint based on deep learning robust to scale changes based on information change across receptive fields includes a feature extractor for extracting a feature from an input image based on a pre-trained deep learning neural network, an information accumulation pyramid module for outputting, from the feature, at least two filter responses corresponding to receptive fields having different scales, an information change detection module for calculating an information change between the at least two filter responses, a keypoint detection module for creating a score map having a keypoint probability of each pixel based on the information change, and a continuous scale estimation module for estimating a scale of a receptive field having a biggest information change for each pixel.

COMPUTER-IMPLEMENTED SEGMENTATION AND TRAINING METHOD IN COMPUTED TOMOGRAPHY PERFUSION, SEGMENTATION AND TRAINING SYSTEM, COMPUTER PROGRAM AND ELECTRONICALLY READABLE STORAGE MEDIUM

A computer-implemented segmentation method for segmenting a core and a penumbra in a four-dimensional computed tomography perfusion dataset of ischemic tissue in an image region of a patient, includes determining at least one parameter map for at least one perfusion parameter from the computed tomography perfusion dataset; and using the at least one parameter map and the computed tomograph perfusion dataset as input data to a trained function to determine output data, the output data including segmentation information of the penumbra and the core.

METHOD FOR PROCESSING IMAGE, METHOD FOR TRAINING FACE RECOGNITION MODEL, APPARATUS AND DEVICE

A method for processing an image includes: obtaining a face image to be processed, and dividing the face image to be processed into image patches; determining respective importance information of the image patches of the face image to be processed; obtaining a pruning rate of a preset vision transformer (ViT) model; inputting the image patches into the ViT model, and pruning inputs of network layers of the ViT model according to the pruning rate and the respective importance information of the image patches, to obtain a result outputted by the ViT model; and determining feature vectors of the face image to be processed according to the result outputted by the ViT model.

Method for glass detection in real scenes

The invention discloses a method for glass detection in a real scene, which belongs to the field of object detection. The present invention designs a combination method based on LCFI blocks to effectively integrate context features of different scales. Finally, multiple LCFI combination blocks are embedded into the glass detection network GDNet to obtain large-scale context features of different levels, thereby realize reliable and accurate glass detection in various scenarios. The glass detection network GDNet in the present invention can effectively predict the true area of glass in different scenes through this method of fusing context features of different scales, successfully detect glass with different sizes, and effectively handle with glass in different scenes. GDNet has strong adaptability to the various glass area sizes of the images in the glass detection dataset, and has the highest accuracy in the field of the same type of object detection.

Image Classification Method And Apparatus

This application relates to an image recognition technology in the field of computer vision of artificial intelligence, and provides an image classification method and apparatus. The method includes: obtaining an input feature map of a to-be-processed image; performing feature extraction processing on the input feature map based on a feature extraction kernel of a neural network, to obtain an output feature map, where each of a plurality of output sub-feature maps is determined based on the corresponding input sub-feature map and the feature extraction kernel, at least one of the output sub-feature maps is determined based on a target matrix obtained after an absolute value is taken, and a difference between the target matrix and the input sub-feature map corresponding to the target matrix is the feature extraction kernel; and classifying the to-be-processed image based on the output feature map, to obtain a classification result of the to-be-processed image.

APPLYING A CONTINUOUS EFFECT VIA MODEL-ESTIMATED CLASS EMBEDDINGS

There is provided methods, devices and techniques to process an image using a deep learning model to achieve continuous effect simulation by a unified network where a simple (effect class) estimator is embedded into a regular encoder-decoder architecture. The estimator allows learning of model-estimated class embeddings of all effect classes (e.g. progressive degrees of the effect), thus representing the continuous effect information without manual efforts in selecting proper anchor effect groups. In an embodiment, given a target age class, there is derived a personalized age embedding which considers two aspects of face aging: 1) a personalized residual age embedding at a model-estimated age of the subject, preserving the subject's aging information; and 2) exemplar-face aging basis at the target age, encoding the shared aging patterns among the entire population. Training and runtime (inference time) embodiments are described including an AR application that generates recommendations ...

TRAINING DATA GENERATION FOR ARTIFICIAL INTELLIGENCE-BASED SEQUENCING

LEARNING OBSERVATION REPRESENTATIONS BY PREDICTING THE FUTURE IN LATENT SPACE

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an encoder neural network that is configured to process an input observation to generate a latent representation of the input observation. In one aspect, a method includes: obtaining a sequence of observations; for each observation in the sequence of observations, processing the observation using the encoder neural network to generate a latent representation of the observation; for each of one or more given observations in the sequence of observations: generating a context latent representation of the given observation; and generating, from the context latent representation of the given observation, a respective estimate of the latent representations of one or more particular observations that are after the given observation in the sequence of observations.

ASSISTING MEDICAL PROCEDURES WITH LUMINESCENCE IMAGES PROCESSED IN LIMITED INFORMATIVE REGIONS IDENTIFIED IN CORRESPONDING AUXILIARY IMAGES

A solution is proposed for assisting a medical procedure. A corresponding method comprises acquiring a luminescence image (205F), based on a luminescence light, and an auxiliary image (205R), based on an auxiliary light different from this luminescence light, of a field of view (103); the field of view (103) contains a region of interest comprising a target body of the medical procedure (containing a luminescence substance) and one or more foreign objects. An auxiliary informative region (210Ri) representative of the region of interest without the foreign objects is identified in the auxiliary image (205R) according to its content, and a luminescence informative region (210Fi) is identified in the luminescence image (205F) according to the auxiliary informative region (210Ri). The luminescence image (205F) is processed limited to the luminescence informative region (210Fi) for facilitating an identification of a representation of the target body therein. A computer program and a corresponding ...

SYSTEMS AND METHODS FOR DETECTION OF CONCEALED THREATS

Described herein are systems for detecting a representation of an object in a radio frequency (RF) image. The system transmits one or more first RF signals toward an object, and receives one or more second RF signals, associated with the one or more transmitted RF signals, that have been reflected from the object. The system determines a plurality of first feature maps corresponding to a RF image associated with the one or more second RF signals. The system combines the plurality of first feature maps. The system further detects a representation of the object in the RF image based at least in part on the combined plurality of first feature maps.

SCANNING ELECTRON MICROSCOPE DEVICE, SEMICONDUCTOR MANUFACTURING DEVICE, AND METHOD OF CONTROLLING SEMICONDUCTOR MANUFACTURING DEVICE

A scanning electron microscope (SEM) device includes: an electron beam source configured to emit an electron beam; a lens unit disposed between the electron beam source and a stage configured to seat an object including structures having a pattern is seated, and including a scanning coil, the scanning coil configured to generate an electromagnetic field to provide a lens, and an astigmatism adjuster; and a control unit. The control unit is configured to change a working distance between the lens unit and the object to obtain a plurality of original images, obtain a pattern image, in which the structures appear, and a plurality of kernel images, in which a distribution of the electron beam on the object appears, from the plurality of original images, and control the astigmatism adjuster to adjust the focus and the astigmatism of the lens unit using feature values extracted from the plurality of kernel images.

SYSTEMS AND METHODS FOR DETECTING OBJECTS

The techniques described herein relate to computerized methods and apparatuses for detecting objects in an image. The techniques described herein further relate to computerized methods and apparatuses for detecting one or more objects using a pre-trained machine learning model and one or more other machine learning models that can be trained in a field training process. The pre-trained machine learning model may be a deep machine learning model.

METHOD OF GENERATING INFERENCE MODEL AND INFORMATION PROCESSING APPARATUS

A computer acquires training data, in which first image data, object information indicating first objects included in the first image data, and relationship information indicating a first relationship between the first objects are associated. The computer executes machine learning that trains, based on the training data, an inference model that infers both second objects included in second image data and a second relationship between the second objects or selectively infers one of the second objects and the second relationship according to an input of the second image data to the inference model. The machine learning uses a penalty term when calculating an error between an inference result of the inference model and the training data. The penalty term causes the error to increase as an overlap between inferred image regions, which are inferred to be image regions in which objects are present in the inference result, increases.

Digital image analysis utilizing multiple human labels

Systems and methods for implementing a multi-label image recognition framework for classifying digital images are provided. The provided multi-label image recognition framework utilizes an iterative, multiple analysis path approach to model training and image classification tasks. A first iteration of the multi-label image recognition framework generates confidence maps for each label, which are shared by the multiple analysis paths to update the confidence maps in subsequent iterations. The provided multi-label image recognition framework permits model training and image classification tasks to be performed more accurately than conventional single-label image recognition frameworks.

Systems and methods to improve feature generation in object recognition

Present embodiments contemplate systems, apparatus, and methods to improve feature generation for object recognition. Particularly, present embodiments contemplate excluding and/or modifying portions of images corresponding to dispersed pixel distributions. By excluding and/or modifying these regions within the feature generation process, fewer unfavorable features are generated and computation resources may be more efficiently employed.

Automatic recognition of images

Presented is a method of automatically performing an action, based on graphical input. The method comprises: receiving, for a user, an input image; comparing the input image with the contents of a user-customized database comprising a plurality of records, each record representing a predefined class of image, wherein the user has previously associated records in the database with respective specified actions; attempting to recognize the image, based on the similarity of the input image to one of the predefined classes of image represented in the user-customised database; and if the image is recognized, performing the action previously associated by the user with the class. Also presented is apparatus for recognizing an image and a method of constructing a user-customized database.

Method and apparatus of compiling image database for three-dimensional object recognition

A method of compiling an image database for a three-dimensional object recognition including the steps of: when a plurality of images each showing an object from different viewpoint are inputted, extracting local features from each of the images, and expressing the local features using feature vectors; forming sets of the feature vectors, each set representing a same part of the object from a series of the viewpoints, and generating subspaces, each subspace representing a characteristic of each set; and storing each subspace to the image database with an identifier of the object to perform a recognition process that is realized by the steps of: when at least one image of an object is given as a query, extracting query feature vectors; determining the subspace most similar to each query feature vector; and executing a counting process to the identifiers to retrieve an object most similar to the query.

Region segmented image data creating system and feature extracting system for histopathological images

A region segmented image data creating system for histopathological images is provided. The region segmented image data creating system is capable of creating region segmented image data required to generating a region segmented image. A first bi-level image data creating section 12 creates first bi-level image data, in which nucleus regions can be discriminated from other regions, from histopathological image data. A second bi-level image data creating section 14 creates second bi-level image data, in which a background regions can be discriminated from other regions, from the histopathological image data. A three-level image data creating section 15 clarifies cytoplasm regions by computing a negative logical addition of the first bi-level image data and the second bi-level image data, and to create three-level image data as the region segmented image data.

Spectral image dimensionality reduction system and method

Methods for reducing dimensionality of hyperspectral image data having a number of spatial pixels, each associated with a number of spectral dimensions, include receiving sets of coefficients associated with each pixel of the hyperspectral image data, a set of basis vectors utilized to generate the sets of coefficients, and either a maximum error value or a maximum data size. The methods also include calculating, using a processor, a first set of errors for each pixel associated with the set of basis vectors, and one or more additional sets of errors for each pixel associated with one or more subsets of the set of basis vectors. Utilizing such errors calculations, an optimum size of the set of basis vectors may be ascertained, allowing for either a minimum amount of error within the maximum data size, or a minimum data size within the maximum error value.

Face recognition using multilayered discriminant analysis

Face recognition using multilayered discriminant analysis includes systems and methods applying an initial linear discriminate analysis to a database of face images in a more-or less conventional manner. Initial fuzzy logic then is applied to the results of the initial linear discriminate analysis to produce a subset of the database of face images. Thereafter, a subsequent linear discriminate analysis is applied to the subset of the database of face images and subsequent fuzzy logic is applied to the results of the subsequent linear discriminate analysis to produce a further subset of the subset of the database of face images. The application of the subsequent linear discriminate analysis and application of the subsequent fuzzy logic may be repeated until the further subset contains only one, or zero, face images.

Method for closed-loop controlling a laser processing operation and laser material processing head using the same

The present invention relates to a method for closed-loop controlling a processing operation of a workpiece, comprising the steps of: (a) recording a pixel image at an initial time point of an interaction zone by means of a camera, wherein the workpiece is processed using an actuator having an initial actuator value; (b) converting the pixel image into a pixel vector; (c) representing the pixel vector by a sum of predetermined pixel mappings each multiplied by a corresponding feature value; (d) classifying the set of feature values on the basis of learned feature values into at least two classes of a group of classes comprising a first class of a too high actuator value, a second class of a sufficient actuator value and a third class of a too low actuator value at the initial time point; (e) performing a control step for adapting the actuator value by minimizing the error e t between a quality indicator y e and a desired value; and (f) repeating the steps (a) to (e) for further time points to perform a closed-loop controlled processing operation.

Image segmentation based on approximation of segmentation similarity

A system and a method for image segmentation use segmentation maps of one or more similar images as a basis for the segmentation. The method includes generating an image signature for an input image to be segmented and identifying at least one similar image from a set of images, based on the image signature of the input image and image signatures of images in the set of images. The similarity may be computed after first projecting the image signatures into a feature space where similarity is more likely to agree with segmentation map similarity. The input image is segmented, based on the segmentation map of one or more of the at least one identified similar images.

Method for Representing Images Using Quantized Embeddings of Scale-Invariant Image Features

Scale-invariant features are extracted from an image. The features are projected to a lower dimensional random projection matrix by multiplying the features by a matrix of random entries. The matrix of random projections is quantized to produce a matrix of quantization indices, which form a query vector for searching a database of images to retrieve metadata related to the image.

Region refocusing for data-driven object localization

A system and method are provided for segmenting an image. The method includes computing an image signature for an input image. One or more similar images are identified from a first set of images, based on the image signature of the input image and image signatures of images in the first set of images. The similar image or images are used to define a cropped region of the input image and a second image signature is computed, this time for the cropped region. One or more similar images are identified from a second set of images, based on the cropped image signature and the image signatures of images in the second set of images. The input image is segmented based on a segmentation map of at least one of the similar images identified in the second set of images.

Systems and methods for quantum processing of data

Systems, methods and aspects, and embodiments thereof relate to unsupervised or semi-supervised features learning using a quantum processor. To achieve unsupervised or semi-supervised features learning, the quantum processor is programmed to achieve Hierarchal Deep Learning (referred to as HDL) over one or more data sets. Systems and methods search for, parse, and detect maximally repeating patterns in one or more data sets or across data or data sets. Embodiments and aspects regard using sparse coding to detect maximally repeating patterns in or across data. Examples of sparse coding include L0 and L1 sparse coding. Some implementations may involve appending, incorporating or attaching labels to dictionary elements, or constituent elements of one or more dictionaries. There may be a logical association between label and the element labeled such that the process of unsupervised or semi-supervised feature learning spans both the elements and the incorporated, attached or appended label.

System and Method for Free Space Estimation

A system and method for estimating free space including applying a machine learning model to camera images of a navigation area, where the navigation area is broken into cells, synchronizing point cloud data from the navigation area with the processed camera images, and associating probabilities that the cell is occupied and object classifications of objects that could occupy the cells with cells in the navigation area based on sensor data, sensor noise, and the machine learning model.

Superpixel classification method based on semi-supervised k-svd and multiscale sparse representation

The present invention discloses a superpixel classification method based on semi-supervised K-SVD and multiscale sparse representation. The method includes carrying out semi-supervised K-SVD dictionary learning on the training samples of a hyperspectral image; using the training samples and the overcomplete dictionary as the input to obtain the multiscale sparse solution of superpixels; and using the obtained sparse representation coefficient matrix and overcomplete dictionary to obtain the result of superpixel classification by residual method and superpixel voting mechanism. The proposing of the present invention is of great significance to solving the problem of salt and pepper noise and the problem of high dimension and small samples in the field of hyperspectral image classification, as well as the problem of how to effectively use space information in classification algorithm based on sparse representation.

Systems and methods for real-time complex character animations and interactivity

Systems, methods, and non-transitory computer-readable media can identify a virtual deformable geometric model to be animated in a real-time immersive environment. The virtual deformable geometric model comprises a virtual model mesh comprising a plurality of vertices, a plurality of edges, and a plurality of faces. The virtual model mesh is iteratively refined in one or more iterations to generate a refined mesh. Each iteration of the one or more iterations increases the number of vertices, the number of edges, and/or the number of faces. The refined mesh is presented during real-time animation of the virtual deformable geometric model within the real-time immersive environment.

Systems and methods for real-time complex character animations and interactivity

Systems, methods, and non-transitory computer-readable media can receive virtual model information associated with a virtual deformable geometric model. The virtual model information comprises a complex rig comprising a plurality of transforms and a first plurality of vertices defined by a default model, and a simplified rig comprising a second plurality of transforms and a second plurality of vertices. The second plurality of vertices correspond to the first plurality of vertices defined by the default model. The simplified rig and the complex rig are deformed based on an animation to be applied to the virtual deformable geometric model. A set of offset data is calculated. The set of offset data comprises, for each vertex in the first plurality of vertices, an offset between the vertex and a corresponding vertex in the second plurality of vertices.

Pose space dimensionality reduction for pose space deformation of a virtual character

Systems and methods for reducing pose space dimensionality. A plurality of example poses can define an input pose space. Each of the example poses can include a set of joint rotations for a virtual character. The joint rotations can be expressed with a singularity-free mathematical representation. The plurality of example poses can then be clustered into one or more clusters. A representative pose can be determined for each cluster. An output pose space with a reduced dimensionality, as compared to the input pose space, can then be provided.

Face recognition in big data ecosystem using multiple recognition models

A system trains a facial recognition modeling system using an extremely large data set of facial images, by distributing a plurality of facial recognition models across a plurality of nodes within the facial recognition modeling system. The system optimizes a facial matching accuracy of the facial recognition modeling system by increasing a facial image set variance among the plurality of facial recognition models. The system selectively matches each facial image within the extremely large data set of facial images with at least one of the plurality of facial recognition models. The system reduces the time associated with training the facial recognition modeling system by load balancing the extremely large data set of facial images across the plurality of facial recognition models while improving the facial matching accuracy associated with each of the plurality of facial recognition models.

Computer vision transaction monitoring

A machine-learning algorithm is trained on images with a set of diverse items to produce as output feature vectors in a feature-vector space derived for the set. New item images for new items are passed to the algorithm and new feature vectors are projected into the vector space. A classifier for each new item is trained on the new feature vectors to determine whether the new item is new item or is not that new item. During a transaction, an item code scanned for an item and an item image are obtained. The item image is passed to the algorithm, a feature vector is obtained, a corresponding classifier for the item code is retrieved, the feature vector is passed to the classifier, and a determination is provided as to whether the item image and item code matches a specific item that should be associated with the item code.

Product positioning method

A product positioning method includes: collecting a product picture; performing integral image calculation on the product picture; and acquiring, according to the calculated integral image, coordinates of each vertex in the product picture by means of differential calculation. According to the present application, an integral image algorithm is applied to product positioning, such that when the product picture quality is not high, for example, the picture is blurry, and it is thus not convenient to position a product by using a picture edge algorithm or a template matching algorithm, the product picture and a background region can be quickly divided by using the integral image algorithm, thereby positioning the product and not being limited by poor picture quality.

Enforced sparsity for classification

An apparatus for classifying an input includes a classifier and a feature extractor. The feature extractor is configured to generate a feature vector based on the input. The feature vector is also configured to set a number of elements of the feature vector to zero to produce a sparse feature vector. The sparse feature vector has the same dimensions as the feature vector generated by the feature extractor. However, the sparse feature vector includes fewer non-zero elements than the feature vector generated by the feature extractor. The feature vector is further configured to forward the sparse feature vector to the classifier to classify the input.

An apparatus and a method for performing a data driven pairwise registration of three-dimensional point clouds

A method and apparatus for performing a data driven pairwise registration of 3D point clouds, which includes at least one scanner adapted to capture a first local point cloud in a first scan and a second local point cloud in a second scan; a PPF deriving unit adapted to process both captured local point clouds to derive associated point pair features; a PPF-Autoencoder adapted to process the derived point pair features to extract corresponding PPF-feature vectors; a PC-Autoencoder adapted to process the captured local point clouds to extract corresponding PC-feature vectors; a subtracter adapted to subtract the PPF-feature vectors from the corresponding PC-vectors to calculate latent difference vectors for both captured point clouds concatenated to a latent difference vector; and a pose prediction network adapted to calculate a relative pose prediction, between the first and second scan performed by the scanner on the basis of the concatenated latent difference vector.

APPLYING SELF-CONFIDENCE IN MULTI-LABEL CLASSIFICATION TO MODEL TRAINING

A computer model is trained to classify regions of a space (e.g., a pixel of an image or a voxel of a point cloud) according to a multi-label classification. To improve the model's accuracy, the model's self-confidence is determined with respect to its own predictions of regions in a training space. The self-confidence is determined based on the class predictions, such as a difference between the highest-predicted class and a second-highest-predicted class. When these are similar, it may reflect areas for potential improvement by focusing training on these low-confidence areas. Additional training may be performed by including modified training data in subsequent training iterations that focuses on low-confidence areas. As another example, additional training may be performed using the self-confidence to modify a classification loss used to refine parameters of the model. 1. A method for improving computer model training with model-determined confidence scores comprising:training a computer model for an initial training period with an initial training set, the computer model trained to predict, for a region of a space, a plurality of class predictions;identifying a training space having a plurality of regions;for each region in the plurality of regions, applying the computer model to the region to generate a plurality of class predictions; and determining a confidence score for the region based on the plurality of class predictions for the region; andtraining the computer model for a further training period based on the confidence scores for the plurality of regions.2. The method of claim 1 , wherein training the computer model for another training period comprises:grouping the plurality of regions into region subsets corresponding to known classifications for the plurality of regions;for each region subset, determining a ratio describing a proportion of the subset having a confidence score below a threshold;generating a modified space from the space by comparing ...

Distance Metric Learning Using Proxies

The present disclosure provides systems and methods that enable distance metric learning using proxies. A machine-learned distance model can be trained in a proxy space in which a loss function compares an embedding provided for an anchor data point of a training dataset to a positive proxy and one or more negative proxies, where each of the positive proxy and the one or more negative proxies serve as a proxy for two or more data points included in the training dataset. Thus, each proxy can approximate a number of data points, enabling faster convergence. According to another aspect, the proxies of the proxy space can themselves be learned parameters, such that the proxies and the model are trained jointly. Thus, the present disclosure enables faster convergence (e.g., reduced training time). The present disclosure provides example experiments which demonstrate a new state of the art on several popular training datasets.

Mobile terminal and method for generating control command using marker attached to finger

A mobile terminal 200 generating a control command using a marker attached to a finger includes a camera 210 configured to receive a target image including a finger region, a fingertip region detection module 220 configured to detect a finger region in the target image using an image processing technique, and a fingertip region in the finger region, a marker region detection module 230 configured to detect a color marker region attached to the finger in the target image using a color detection technique, a control module 240 configured to operate the fingertip region detection module or the marker region detection module, based on an input from a user, and a coordinate calculation module 250 configured to calculate three-dimensional coordinate values using the fingertip region or the color marker region.

COLLATION APPARATUS, COLLATION METHOD, AND COMPUTER READABLE RECORDING MEDIUM

A collation apparatus includes: a vector-type arithmetic unit that calculates first similarity degrees using first feature points extracted from a target biological image and second feature points in a plurality of registered biological images, and narrows down the registered biological images based on the calculated first similarity degrees; and an arithmetic unit , other than the vector-type arithmetic unit , that calculates second similarity degrees using third feature points extracted from the target biological image and fourth feature points in the registered biological images obtained by the narrowing-down, and specifies a registered biological image based on the calculated second similarity degrees. 1. A collation apparatus comprising:a vector-type arithmetic unit configured to calculate first similarity degrees using first feature points extracted from a target biological image and second feature points in a plurality of registered biological images, and narrow down the registered biological images based on the calculated first similarity degrees; andan arithmetic unit, other than the vector-type arithmetic unit, configured to calculate second similarity degrees using third feature points extracted from the target biological image and fourth feature points in the registered biological images obtained by the narrowing-down, and specify a registered biological image based on the calculated second similarity degrees.2. The collation apparatus according to claim 1 , further comprisinga feature point adjustment unit configured to adjust the number of first feature points.3. The collation apparatus according to claim 2 , whereinthe arithmetic unit includes the feature point adjustment unit.4. The collation apparatus according to claim 1 , wherein the collation apparatus is used for biometric authentication.5. A collation method comprising:using a vector-type arithmetic unit to calculate first similarity degrees using first feature points extracted from a target ...

On-bed monitoring system for range of motion exercises with a pressure sensitive bed sheet

A system includes a pressure sensitive material that provides an indication of applied pressure for multiple locations on the material, and an analysis device in communication with the pressure sensitive material. The analysis device receives the indication of applied pressure, determines, for each of multiple measurement periods, a pressure image from the indication of applied pressure such that a sequence of pressure images is determined, and constructs a manifold representing the sequence of pressure images.

Multimodality image processing techniques for training image data generation and usage thereof for developing mono-modality image inferencing models

Techniques are described for generating mono-modality training image data from multi-modality image data and using the mono-modality training image data to train and develop mono-modality image inferencing models. A method embodiment comprises generating, by a system comprising a processor, a synthetic 2D image from a 3D image of a first capture modality, wherein the synthetic 2D image corresponds to a 2D version of the 3D image in a second capture modality, and wherein the 3D image and the synthetic 2D image depict a same anatomical region of a same patient. The method further comprises transferring, by the system, ground truth data for the 3D image to the synthetic 2D image. In some embodiments, the method further comprises employing the synthetic 2D image to facilitate transfer of the ground truth data to a native 2D image captured of the same anatomical region of the same patient using the second capture modality.

Method for processing electronic data

A method for processing electronic data includes the steps of transforming the electronic data to a matrix representation including a plurality of matrices; decomposing the matrix representation into a series of matrix approximations; and processing, with an approximation process, the plurality of matrices thereby obtaining a low-rank approximation of the plurality of matrices.

OBJECT DETECTION AND DETECTION CONFIDENCE SUITABLE FOR AUTONOMOUS DRIVING

In various examples, detected object data representative of locations of detected objects in a field of view may be determined. One or more clusters of the detected objects may be generated based at least in part on the locations and features of the cluster may be determined for use as inputs to a machine learning model(s). A confidence score, computed by the machine learning model(s) based at least in part on the inputs, may be received, where the confidence score may be representative of a probability that the cluster corresponds to an object depicted at least partially in the field of view. Further examples provide approaches for determining ground truth data for training object detectors, such as for determining coverage values for ground truth objects using associated shapes, and for determining soft coverage values for ground truth objects. 1. A method comprising:determining a region corresponding to an object depicted in a training image for one or more machine learning models (MLMs);assigning coverage values to spatial element regions corresponding to the training image based at least on the spatial element regions at least partially falling within the region; andtraining the one or more MLMs to infer the coverage values in association with detecting the object in the spatial element regions.2. The method of claim 1 , wherein the assigning of the coverage values includes computing a size of a shape at least partially within the region based at least on a dimension of the region claim 1 , wherein the coverage values are assigned to the spatial element regions based at least on the spatial element regions each corresponding to at least a portion of the shape.3. The method of claim 1 , wherein the coverage values are assigned to the spatial element regions based at least on the spatial element regions each including at least a portion of an ellipse claim 1 , a rectangle claim 1 , a circle claim 1 , or a super-ellipse associated with the region.4. The method of ...

Method, apparatus, and non-transitory computer-readable storage medium for verification process

A method for a verification process that performs neighbor discovery for one or more feature points projected to an m-dimensional space (m is a natural number equal to or greater than 2), includes: acquiring a feature point group including one or more feature points projected to coordinate values of the m-dimensional space ordered in a coordinate value order on each of two or more coordinate axes that define the m-dimensional spacer (m is a natural number equal to or greater than 2); selecting a datum axis on which a comparison time number in neighbor discovery is small, the comparison time number being obtained by performing simulation of neighbor discovery.

Image classification by brain computer interface

A method of classifying an image is disclosed. The method comprises: applying a computer vision procedure to the image to detect therein candidate image regions suspected as being occupied by a target; presenting to an observer each candidate image region as a visual stimulus, while collecting neurophysiological signals from a brain of the observer; processing the neurophysiological signals to identify a neurophysiological event indicative of a detection of the target by the observer; and determining an existence of the target in the image is based, at least in part, on the identification of the neurophysiological event.

Automated methods and systems for locating document subimages in images to facilitate extraction of information from the located document subimages

The present document is directed to methods and subsystems that identify and characterize document-containing subimages in a document-containing image. In one implementation, each type of document is modeled as a set of features that are extracted from a set of images known to contain the document. To locate and characterize a document subimage in an image, the currently described methods and subsystems extract features from the image and then match model features of each model in a set of models to the extracted features to select the model that best corresponds to the extracted features. Additional information contained in the selected model is then used to identify the location of the subimage corresponding to the document and to process the document subimage to correct for a variety of distortions and deficiencies in order to facilitate subsequent data extraction from the corrected document subimage.

An arrangement for producing head related transfer function filters

When three-dimensional audio is produced by using headphones particular HRTF-filters are used the sound for left and right channels of the headphone. As the morphology of every ear is different, it is beneficial to have HRTF-filters particularly designed for the user of headphones. Such filters may be produced deriving ear geometry from a plurality of images taken with an ordinary camera, detecting necessary features from images and fitting said features to a model that has been produced from accurately scanned ears comprising representative values for different sizes and shapes. Taken images are sent to a server that performs the necessary computations and submits the data further or produces the requested filter.

Model training method and apparatus, keypoint positioning method and apparatus, device and medium

Provided are a training method and apparatus for a human keypoint positioning model, a human keypoint positioning method and apparatus, a device, a medium and a program product. The training method includes determining an initial positioned point of each of keypoints; acquiring N candidate points of each keypoint according to a position of the initial positioned point; extracting a first feature image, and forming N sets of graph structure feature data according to the first feature image and the N candidate points; performing graph convolution on the N sets of graph structure feature data to obtain N sets of offsets; correcting initial positioned points of all the keypoints to obtain N sets of current positioning results; and calculating each set of loss values according to labeled true values of all the keypoints and each set of current positioning results, and performing supervised training on the positioning model.

LEARNING APPARATUS, LEARNING METHOD, AND LEARNING PROGRAM, REGION-OF-INTEREST EXTRACTION APPARATUS, REGION-OF-INTEREST EXTRACTION METHOD, AND REGION-OF-INTEREST EXTRACTION PROGRAM, AND LEARNED EXTRACTION MODEL

An extraction model is constituted of an encoder that extracts a feature amount of a first image of a first representation format to derive a feature map of the first image, a first decoder that derives a second virtual image of a second representation format different from the representation format of the first image on the basis of the feature map, a first discriminator that discriminates a representation format of an input image and whether the input image is a real image or a virtual image, and outputs a first discrimination result, a second decoder that extracts a region of interest of the first image on the basis of the feature map, and a second discriminator that discriminates whether an extraction result of the region of interest by the second decoder is an extraction result of a first image with ground-truth mask or an extraction result of a first image without ground-truth mask, and outputs a second discrimination result.

AUTOMATIC TOPOLOGY MAPPING PROCESSING METHOD AND SYSTEM BASED ON OMNIDIRECTIONAL IMAGE INFORMATION

An automatic topology mapping processing method and system. The automatic topology mapping processing method includes the steps of: obtaining, by the automatic topology mapping processing system, a plurality of images, wherein at least two of the plurality of images include a common area in which a common space is captured; extracting, by the automatic topology mapping processing system, from respective images, features of the respective images through a feature extractor using a neural network; and determining, by the automatic topology mapping processing system, mapping images of the respective images on the basis of the features extracted from the respective images.

Biomarker Method

Provided herein is a method for quantifying the spatial distribution of cells or sub-cellular structures comprising (a) receiving image data comprising a plurality of biomarkers, N, wherein the data represents a spatial map of the biomarkers; (b) processing the data to obtain a set of coordinates, wherein each coordinate denotes the location of a cell or sub-cellular structure represented by a biomarker or combination of biomarkers; and (c) processing the set of coordinates into a two-dimensional symmetric (2 N −1)×(2 N −1) or 2 N ×2 N matrix (D). Also provided is the use of this method for assigning subjects one or more clinical characteristics, the use of this method for forming and/or testing scientific hypotheses for one or more interventions, and apparatus comprising at least one processor configured to perform the method.

System And Method For Detection Of High-Interest Events In Video Data

A method for event identification in video data includes identifying a feature vector having data corresponding to at least one of a position and a direction of movement of an object in video data, generating an estimated feature vector corresponding to the feature vector using a dictionary including a plurality of basis vectors, identifying an error between the estimated feature vector and the feature vector, identifying a high-interest event in the video data in response to the identified error exceeding a threshold, and displaying the video data including the high-interest event on a video output device only in response to the error exceeding the threshold.

METHODS AND SYSTEMS FOR AUTOMATICALLY MATCHING AUDIO CONTENT WITH VISUAL INPUT

A method and system for automatically matching audio content with visual input by obtaining a video or image(s) from a digital camera, analyzing the video or image(s) to extract image label(s), mapping the image label(s) to predefined input-output relationships to determine a set of input feature values, and selecting a set of music tracks from a plurality of music tracks stored on a storage device having associated semantic feature values most closely matching the determined input feature values to create a playlist. 1. A computer-implemented method for automatically matching audio content with visual input , the method comprising:providing a storage device comprising a plurality of music tracks, each music track having linked therewith a feature vector comprising a set of semantic feature values;defining input-output relationships between a set of labels and a corresponding set of semantic feature values for each label;obtaining at least one image from a digital camera;analyzing the at least one image to extract at least one image label describing the visual content of the at least one image;mapping the at least one image label to the input-output relationships to determine a set of input feature values;calculating an input feature vector based on the set of input feature values; andselecting a set of music tracks from the plurality of music tracks based on calculated vector distances between the input feature vector and the respective linked feature vectors, the selected set of music tracks having associated semantic feature values most closely matching the input feature values; andcreating a playlist for the at least one image comprising the set of music tracks.2. The method according to claim 1 , wherein the input-output relationships are defined by providing a semantic matrix defining relationships between a set of labels and a corresponding set of semantic features claim 1 , wherein the values of the semantic matrix represent a relevance of each semantic ...

TRAINING METHOD AND APPARATUS FOR TARGET DETECTION MODEL, DEVICE AND STORAGE MEDIUM

Provided are a training method and apparatus for a target detection model, a device and a storage medium. The training method is described below. A feature map of a sample image is processed through a classification network of an initial model and a heat map and a classification prediction result of the feature map are obtained, a classification loss value is determined according to the classification prediction result and classification supervision data of the sample image, and a category probability of pixels in the feature map is determined according to the heat map of the feature map and a probability distribution map of the feature map is obtained; the feature map is processed through a regression network of the initial model and a regression prediction result is obtained, and a regression loss value is determined. 1. A training method for a target detection model , comprising:processing, through a classification network of an initial model, a feature map of a sample image and obtaining a heat map and a classification prediction result of the feature map, determining a classification loss value according to the classification prediction result and classification supervision data of the sample image, and determining, according to the heat map of the feature map, a category probability of pixels in the feature map and obtaining a probability distribution map of the feature map;processing, through a regression network of the initial model, the feature map and obtaining a regression prediction result, and determining a regression loss value according to the probability distribution map, the regression prediction result and regression supervision data of the sample image; andtraining the initial model according to the regression loss value and the classification loss value, and obtaining the target detection model.2. The method according to claim 1 , wherein processing claim 1 , through the classification network of the initial model claim 1 , the feature map and ...

Long-tail large scale face recognition by non-linear feature level domain adaption

A computer-implemented method, system, and computer program product are provided for facial recognition. The method includes receiving, by a processor device, a plurality of images. The method also includes extracting, by the processor device with a feature extractor utilizing a convolutional neural network (CNN) with an enlarged intra-class variance of long-tail classes, feature vectors for each of the plurality of images. The method additionally includes generating, by the processor device with a feature generator, discriminative feature vectors for each of the feature vectors. The method further includes classifying, by the processor device utilizing a fully connected classifier, an identity from the discriminative feature vector. The method also includes control an operation of a processor-based machine to react in accordance with the identity.

METHOD FOR GLASS DETECTION IN REAL SCENES

The invention discloses a method for glass detection in a real scene, which belongs to the field of object detection. The present invention designs a combination method based on LCFI blocks to effectively integrate context features of different scales. Finally, multiple LCFI combination blocks are embedded into the glass detection network GDNet to obtain large-scale context features of different levels, thereby realize reliable and accurate glass detection in various scenarios. The glass detection network GDNet in the present invention can effectively predict the true area of glass in different scenes through this method of fusing context features of different scales, successfully detect glass with different sizes, and effectively handle with glass in different scenes. GDNet has strong adaptability to the various glass area sizes of the images in the glass detection dataset, and has the highest accuracy in the field of the same type of object detection. 1. A method for glass detection in a real scene , wherein the method specifically includes the following steps:step 1 constructing glass detection dataset GDDusing cameras and smartphones to capture glass images for constructing a glass detection dataset GDD; the glass detection dataset GDD contains images with different scenes and different sizes of glass scenes to ensure the diversity of network learning and the applicability of the network; the images are taken in real scenes; the captured images in the glass detection dataset GDD are divided into training set and testing set;step 2 multi-level feature extractor extracts featuresinputting the images in the training set of the GDD dataset constructed in step 1 into the multi-level feature extractor MFE to harvest features of different levels; the multi-level feature extractor MFE is implemented by using a feature extraction network;step 3 constructing a large-scale contextual feature integration LCFI blockusing the cross convolution to construct LCFI block: ...

MODEL TRAINING METHOD, CONTENT GENERATION METHOD, AND RELATED APPARATUSES

A processing circuitry obtains a training set based on an interaction process of a plurality of cards. The training set includes a plurality of video frames, and a video frame includes a trigger region for triggering an action during the interaction process. The processing circuitry determines feature regions in the trigger region. A feature region includes a card and is set with an action label for indicating a first training content in the feature region. The processing circuitry also determines a feature vector based on the plurality of video frames. The feature vector indicates a triggering of the feature region. The processing circuit inputs the feature vector into a first model for training to obtain a second model. The first model associates the feature vector with the action label, and the second model is used for indicating a correspondence between a content of the card and the action label. 1. A method for model training in an electronic device , comprising:obtaining a training set based on an interaction process of a plurality of cards, the training set comprising a plurality of video frames, and a video frame in the plurality of video frames comprising a trigger region for triggering an action during the interaction process;determining feature regions in the trigger region of the video frame, a feature region in the feature regions including a card and being set with an action label for indicating a first training content in the feature region;determining a feature vector based on the plurality of video frames, the feature vector indicting a triggering of the feature region; andinputting the feature vector into a first model for training to obtain a second model, the first model being used for associating the feature vector with the action label, and the second model being used for indicating a correspondence between a content of the card and the action label.2. The method according to claim 1 , wherein the determining the feature regions in trigger ...

Method for visual localization and related apparatus

Visual localization method and related apparatus are disclosed. In the method, a first candidate image sequence is determined from image library, the image library being configured to construct electronic map, image frames in the first candidate image sequence being sequentially arranged according to degrees of matching with first image, and the first image being an image collected by a camera; an order of the image frames in the first candidate image sequence is adjusted according to target window to obtain second candidate image sequence, the target window being multiple successive image frames including target image frame and determined from the image library, the target image frame being an image matching with second image, which is collected by the camera before the first image is collected, in the image library; and target posture of the camera when the first image is collected is determined according to the second candidate image sequence.

PEDESTRIAN DETECTION METHOD AND APPARATUS, COMPUTER-READABLE STORAGE MEDIUM, AND CHIP

This application relates to the field of artificial intelligence, and specifically, to the field of computer vision. The method includes: performing feature extraction on an image to obtain a basic feature map of the image; determining a proposal of a region possibly including a pedestrian in the image; processing the basic feature map of the image to obtain an object visibility map in which a response to a pedestrian visible part is greater than a response to a pedestrian blocked part and a background part; performing weighted summation processing on the object visibility map and the basic feature map to obtain an enhanced feature map of the image; and determining, based on the proposal of the image and the enhanced feature map of the image, a bounding box including a pedestrian in the image and a confidence level of the bounding box including the pedestrian in the image. 1. A pedestrian detection method , comprising:obtaining an image;performing feature extraction on the image to obtain a basic feature map of the image;determining, based on the basic feature map, a proposal of the image, wherein the proposal is a bounding box of a region possibly comprising a pedestrian in the image;processing the basic feature map of the image to obtain an object visibility map of the image, wherein a pixel value of a pedestrian visible part in the object visibility map is greater than a pixel value of a pedestrian invisible part in the object visibility map;performing fusion processing on the basic feature map of the image and the object visibility map of the image to obtain an enhanced feature map of the image;determining, based on the proposal of the image and the enhanced feature map of the image, a feature corresponding to the proposal, wherein the feature corresponding to the proposal comprises a region feature of the proposal, and the region feature of the proposal is a feature of a region that is in the enhanced feature map and that is located in the proposal; ...

Detecting multi-object anomalies utilizing a low rank sparsity model

Methods and systems for detecting anomalies in transportation related video footage. In an offline training phase, receiving video footage of a traffic location can be received. Also, in an offline training phase, event encodings can be extracted from the video footage and collected or compiled into a training dictionary. One or more input video sequences captured at the traffic location or a similar traffic location can be received in an online detection phase. Then, an event encoding corresponding to the input video sequence can be extracted. The event encoding can be reconstructed with a low rank sparsity prior model applied with respect to the training dictionary. The reconstruction error between actual and reconstructed event encodings can then be computed in order to determine if an event thereof is anomalous by comparing the reconstruction error with a threshold.

Methods for delineating cellular regions and classifying regions of histopathology and microanatomy

Embodiments disclosed herein provide methods and systems for delineating cell nuclei and classifying regions of histopathology or microanatomy while remaining invariant to batch effects. These systems and methods can include providing a plurality of reference images of histology sections. A first set of basis functions can then be determined from the reference images. Then, the histopathology or microanatomy of the histology sections can be classified by reference to the first set of basis functions, or reference to human engineered features. A second set of basis functions can then be calculated for delineating cell nuclei from the reference images and delineating the nuclear regions of the histology sections based on the second set of basis functions.

Systems and methods for fractal-based visual searching

The present disclosure provides a visual search engine (VSE) configured to perform visual searches. The VSE may receive a query image for searching through a plurality of images stored in a dataset. Different ones of the images stored in the dataset may be indexed or logically grouped together based on a fractal transform that associates images depicting similar content with each other. A fractal transform of the query image may be used to identify a plurality of images from the dataset to be searched based on the query image. A modified image triplet technique using the query image, a derived set of similar images, and a derived set of dissimilar images may be utilized to identify features of the images being searched. Search results logic of the VSE may apply deep learning techniques to the feature sets to identify a set of search results to return for the query image.

Individual identification information generation method, individual identification information generation device, and program

An individual identification information generation device 100 of the present invention includes an extraction unit 121 and a generation unit 122. The extraction unit 121 extracts, on the basis of product information in which manufacturing state information representing a manufacturing state of a product and individual identification information using a surface pattern of the product acquired from a captured image of the product are associated with each other, a relationship between the manufacturing state information and the individual identification information. The generation unit 122 generates, on the basis of the relationship between the manufacturing state information and the individual identification information and of the manufacturing state information of a given product, the individual identification information of the given product.

IMAGE PROCESSING METHOD, APPARATUS, AND DEVICE, AND STORAGE MEDIUM

An image processing method is provided. The image processing method includes: acquiring first second input images; extracting a content feature of the first input image; extracting an attribute feature of the second input image; performing feature fusion and mapping processing on the content feature of the first input image and the attribute feature of the second input image by using a feature transformation network to obtain a target image feature, the target image feature having the content feature of the first input image and the attribute feature of the second input image; and generating an output image based on the target image feature. 1. An image processing method , performed by at least one processor of an image processing device and comprising:acquiring a first input image and a second input image;extracting a content feature of the first input image;extracting an attribute feature of the second input image;performing feature fusion and mapping processing on the content feature of the first input image and the attribute feature of the second input image by using a feature transformation network to obtain a target image feature, the target image feature having the content feature of the first input image and the attribute feature of the second input image; andgenerating an output image based on the target image feature.2. The image processing method according to claim 1 , wherein the extracting the content feature of the first input image comprises: extracting the content feature of the first input image by using a content encoder network claim 1 ,wherein the extracting the attribute feature of the second input image comprises: extracting the attribute feature of the second input image by using an attribute encoder network, andwherein the generating the output image based on the target image feature comprises: generating the output image based on the target image feature by using a decoder network.3. The image processing method according to claim 2 , wherein ...

System and Method for Multimedia Analytic Processing and Display

The present disclosure includes systems and methods for multimedia image analytic including automated binarization, segmentation, and enhancement using bio-inspired based visual morphology schemes. The present disclosure further includes systems and methods for biometric multimedia content authentication using extracted geometric features and one or more of the binarization, segmentation, and enhancement methods.

Systems and methods for image feature extraction

This description relates to image feature extraction. In some examples, a system can include a keypoint detector and a feature list generator. The keypoint detector can be configured to upsample a keypoint score map to produce an upsampled keypoint score map. The keypoint score map can include feature scores indicative of a likelihood of at least one feature being present at keypoints in an image. The feature list generator can be configured to identify a subset of keypoints of the keypoints in the image using the feature scores of the up sampled keypoint score map, determine descriptors for the subset of keypoints based on a feature description map, and generate a keypoint descriptor map for the image based on the determined descriptors.

METHOD AND DEVICE FOR AUTOMATIC IDENTIFICATION OF LABELS OF AN IMAGE

Disclosed herein is a method comprising: determining a first value of a single-label of an image and a first value of a multi-label of the image, based on a feature map of the image; producing a weighted feature map from the feature map based on a characteristic of features of the feature map; determining a second value of the multi-label of the image by performing spatial regularization on the weighted feature map; determining a third value of the multi-label based on the first value of the multi-label and the second value of the multi-label. 1. A computer-implemented method for identifying labels of an image comprising:determining a first value of a single-label of the image and a first value of a multi-label of the image, based on a feature map of the image;producing a weighted feature map from the feature map based on a characteristic of features of the feature map;determining a second value of the multi-label of the image by performing spatial regularization on the weighted feature map;determining a third value of the multi-label based on the first value of the multi-label and the second value of the multi-label.2. The method of claim 1 , wherein the characteristic is correlation of the features with the multi-label.3. The method of claim 2 , wherein the correlation is spatial correlation or sematic correlation.4. The method of claim 1 , wherein the third value of the multi-label is a weighted average of the first value of the multi-label and the second value of the multi-label.5. The method of claim 1 , further comprising determining a fourth value of the multi-label from the third value of the multi-label based on sematic correlation between the single-label and the multi-label.6. The method of claim 5 , further comprising applying a threshold to the fourth value of the multi-label.7. The method of claim 1 , further comprising determining a second value of the single-label from the first value of the single-label based on sematic correlation between the ...

COMPUTING DEVICE AND OPERATING METHOD THEREFOR

A computing device for performing image quality processing of an image, and an operating method thereof are provided. The computing device includes: a memory storing at least one instruction; and a processor configured to execute the at least one instruction stored in the memory, to: extract features of an input image by using a first neural network, recognize at least one instance in the input image from the features of the input image by using a second neural network, obtain an image instance quality score map by predicting a quality score corresponding to each instance of the at least one instance in the input image based on the features of the input image by using a third neural network, and perform image quality processing differently for each instance of the at least one instance in the input image by using the quality score corresponding to each instance of the at least one instance in the input image. 1. A computing device comprising:a memory storing at least one instruction; and extract features of an input image by using a first neural network,', 'recognize at least one instance in the input image from the features of the input image by using a second neural network,', 'obtain an image instance quality score map by predicting a quality score corresponding to each instance of the at least one instance in the input image based on the features of the input image by using a third neural network, and', 'perform image quality processing differently for each instance of the at least one instance in the input image by using the quality score corresponding to each instance of the at least one instance in the input image., 'a processor configured to execute the at least one instruction stored in the memory, to2. The computing device of claim 1 , wherein the image instance quality score map includes position information and a quality score of each instance of the at least one instance in the input image.3. The computing device of claim 1 , wherein the first neural ...

Weakly supervised learning with whole slide images

Techniques are provided for determining classifications based on WSIs. A varied-size feature map is generated for each training WSI by generating a grid of patches for the training WSI, segmenting the training WSI into tissue and non-tissue areas, and converting patches comprising the tissue areas into tensors. Bounding boxes are generated based on the patches comprising tissue areas and segmented into feature map patches. A fixed-size feature map is generated based on a subset of the feature map patches. A classifier model is trained to process fixed-size feature maps corresponding to the training WSIs such that, for each fixed-size feature map, the classifier model is operable to assign a WSI-level tissue or cell morphology classification or regression based on the tensors. A classification engine is configured to use the trained classifier model to determine a WSI-level tissue or cell morphology classification or regression for a test WSI.

Vision Based Target Tracking that Distinguishes Facial Feature Targets

A facial recognition method using online sparse learning includes initializing target position and scale, extracting positive and negative samples, and extracting high-dimensional Haar-like features. A sparse coding function can be used to determine sparse Haar-like features and form a sparse feature matrix, and the sparse feature matrix in turn is used to classify targets.

Pointer recognition for analog instrument image analysis

Systems and processes for identifying a pointer in an image of an analog instrument are provided herein. An instrument contour in the image corresponding to the analog instrument may be identified. A plurality of candidate pointer contours in the image may be identified and screened using one or more geometric property screening techniques including an evaluation of a geometric area, a distance parameter, and/or a gravity center of the plurality of candidate pointer contours. Principal component analysis (PCA) may be performed to select an identified pointer contour from among the reduced plurality of candidate pointer contours. A linear regression model may be applied to pixel points in the contour area of the identified pointer contour and a slope and angle of an associated pointer represented by the identified pointer contour may be determined based on an output of the linear regression model.

Video analysis method and apparatus and computer program

Provided is a video analysis (VA) method comprises calculating, by a VA apparatus, a size of each of a plurality of frames by summing sizes of a plurality of packets constituting each of the plurality of frames, analyzing, by the VA apparatus, the size of each of the plurality of frames constituting a video to determine a pattern of the sizes of the plurality of frames and determining, by the VA apparatus, whether there is a motion in the video based on the size pattern.

COMPENSATION OF ORGAN DEFORMATION FOR MEDICAL IMAGE REGISTRATION

Systems and methods for medical image registration are provided. A first input medical image and a second input medical image of one or more anatomical objects arc received. For each respective anatomical object of the one or more anatomical objects, a region of interest comprising the respective anatomical object is detected in one of the first input medical image or the second input medical image, the region of interest is extracted from the first input medical image and from the second input medical image, and a motion distribution of the respective anatomical object is determined from one of the region of interest extracted from the first input medical image or the region of interest extracted from the second input medical image using a motion model specific to the respective anatomical object. The first input medical image and the second input medical image are registered based on the motion distribution of each respective anatomical object of the one or more anatomical objects to generate a fused image. 1. A method , comprising:receiving a first input medical image and a second input medical image of one or more anatomical objects; detecting a region of interest comprising the respective anatomical object in one of the first input medical image or the second input medical image,', 'extracting the region of interest from the first input medical image and from the second input medical image, and', 'determining a motion distribution of the respective anatomical object from the region of interest extracted from one of the first input medical image or the region of interest extracted from the second input medical image using a motion model specific to the respective anatomical object; and, 'for each respective anatomical object of the one or more anatomical objectsregistering the first input medical image and the second input medical image based on the motion distribution of each respective anatomical object of the one or more anatomical objects.2. The method of ...

MAPPING AND LOCALIZATION SYSTEM FOR AUTONOMOUS VEHICLES

Accurate vehicle localization is arguably the most critical and fundamental task for autonomous vehicle navigation. While dense 3D point-cloud-based maps enable precise localization, they impose significant storage and transmission burdens when used in city-scale environments. A highly compressed representation for LiDAR maps, along with an efficient and robust real-time alignment algorithm for on-vehicle LiDAR scans, is proposed here. The proposed mapping framework, requires less than 0.1% of the storage space of the original 3D point cloud map. In essence, mapping framework emulates an original map through feature likelihood functions. In particular, the mapping framework models planar, pole and curb features. These three feature classes are long-term stable, distinct and common among vehicular roadways. Multiclass feature points are extracted from LiDAR scans through feature detection. A new multiclass-based point-to-distribution alignment method is also proposed to find the association and alignment between the multiclass feature points and the map. 1. A method for constructing a two-dimensional feature map of a scene , comprising:receiving point cloud data for a scene;projecting pixels in the point cloud data into an x-y plane, where the x-y plane is substantially parallel with ground in the scene;dividing the x-y plane into an array of cells;classifying one or more in the array of cells as potential feature candidate cells;generating line segments by connecting cells in the array of cells which are classified as potential feature candidate cells;designating each line segment as a feature in the scene;representing each feature in the scene with an approximation function having a mean and a covariance; andfor each feature, storing the approximation function assigned to a given feature, along with a global coordinate for the given feature, in a non-transitory data store, thereby constructing a feature map of the scene.2. The method of further comprises extracting ...

DRIVE ASSIST APPARATUS, DRIVE ASSIST METHOD, AND DRIVE ASSIST SYSTEM

A drive assist apparatus includes a storage unit that stores a three-dimensional model indicating a moving region, an input unit that receives, from a sensor group installed in the moving region, first height information indicating a first height which is a height of the mobile object and second height information indicating a second height which is a height of an object that satisfies a predetermined distance criterion from the mobile object, an extraction unit that extracts, from the three-dimensional model, a first plan view based on the first height information and a second plan view based on the second height information, a generation unit that generates a combined map for two-dimensionally showing the moving region and assisting the driving of the mobile object by combining the first plan view and the second plan view, and an output unit that transmits a generated combined map to the mobile object. 1. A drive assist apparatus for assisting driving of a mobile object that moves within a moving region , the drive assist apparatus comprising:a storage unit that stores at least a three-dimensional model indicating the moving region;an input unit that is able to receive, from a sensor group installed in the moving region, first height information indicating a first height which is a height of the mobile object and second height information indicating a second height which is a height of an object that satisfies a predetermined distance criterion from the mobile object;an extraction unit that extracts a first plan view from the three-dimensional model based on the first height information and extracts a second plan view from the three-dimensional model based on the second height information;a generation unit that generates a combined map for two-dimensionally showing the moving region and assisting the driving of the mobile object by combining the first plan view and the second plan view; andan output unit that is able to transmit a generated combined map to the ...

Image processing using self-attention

An image processing device for identifying one or more characteristics of an input image, the device including a processor configured to: receive the input image, the input image extending along a first axis and a second axis; form a series of attribute maps based on the received input image; perform a first correlation operation by identifying regions in respect of which the patterns of multiple ones of the series of attribute maps are correlated, and forming a first output in dependence on that operation; perform a second correlation operation for identifying combinations of (i) attributes and (ii) portions of the image having common location in terms of the first axis, and forming a second output in dependence on that operation; and form a representation of the one or more characteristics of the input image in dependence on at least the first output and the second output.

Face recognition method and apparatus, computer device, and storage medium

A face recognition method includes: obtaining a first feature image that describes a face feature of a target face image and a first feature vector corresponding to the first feature image; obtaining a first feature value that represents a degree of difference between a face feature in the first feature image and that in the target face image; obtaining a similarity between the target face image and a template face image according to the first feature vector, the first feature value, and a second feature vector and a second feature value corresponding to a second feature image of the template face image, the second feature value describing a degree of difference between a face feature in the second feature image and that in the template face image; and determining, when the similarity is greater than a preset threshold, that the target face image matches the template face image.

DATA AUGMENTATION BASED ON ATTENTION

Implementations of the present disclosure relate to methods, devices, and computer program products for data augmentation. In the method, mixed data is generated from first data and second data, and the mixed data comprises a first portion from the first data and a second portion from the second data. An attention map is obtained for the mixed data based on distributions of the first and second portions in the mixed data, here the attention map describes contributions of the first and second data to the mixed data. A label is determined for the mixed data based on the attention map and a first label for the first data and a second label for the second data. With these implementations, the label is determined based on the contributions of the first and second images in an accurate and effective way, and thus has a value that is much closer to the ground true. 1. A method for data augmentation , comprising:generating mixed data from first data and second data, the mixed data comprising a first portion from the first data and a second portion from the second data;obtaining an attention map for the mixed data based on distributions of the first and second portions in the mixed data, the attention map describing contributions of the first and second data to the mixed data; anddetermining a label for the mixed data based on the attention map and a first label for the first data and a second label for the second data.2. The method of claim 1 , wherein obtaining the attention map comprises:dividing the mixed data into a plurality of data blocks;determining a plurality of data tokens for the plurality of data blocks, respectively; andobtaining the attention map based on a self-attention operation to a token sequence that comprises the plurality of data tokens and a class token associated with the attention map.3. The method of claim 2 , wherein obtaining the attention map comprises:determining a query parameter and a key parameter for the self-attention operation based on ...

METHOD AND APPARATUS OF TRAINING IMAGE RECOGNITION MODEL, METHOD AND APPARATUS OF RECOGNIZING IMAGE, AND ELECTRONIC DEVICE

The present application provides a method and an apparatus of training an image recognition model, a method and an apparatus of recognizing an image, and an electronic device, which relates to a field of an image processing technology, and in particular to artificial intelligence and computer vision technology. A specific implementation scheme of the present disclosure includes: determining a training sample set including a plurality of sample pictures and a text label for each sample picture; extracting an image feature of each sample picture and a semantic feature of each sample picture based on a feature extraction network of a basic image recognition model; and training the basic image recognition model based on the extracted image feature of each sample picture, the extracted semantic feature of each sample picture, the text label for each sample picture, a predetermined image classification loss function, and a predetermined semantic classification loss function. 1. A method of training an image recognition model , comprising:determining a training sample set comprising a plurality of sample pictures and a text label for each sample picture; wherein at least part of the plurality of sample pictures in the training sample set contains an irregular text, an occluded text or a blurred text;extracting an image feature of each sample picture and a semantic feature of each sample picture based on a feature extraction network of a basic image recognition model; andtraining the basic image recognition model based on the extracted image feature of each sample picture, the extracted semantic feature of each sample picture, the text label for each sample picture, a predetermined image classification loss function, and a predetermined semantic classification loss function.2. The method of claim 1 , wherein the sample picture comprises at least one of a shop sign picture claim 1 , a billboard picture and a slogan picture.3. The method of claim 1 , wherein the training the ...

Intelligent vibration digital twin systems and methods for industrial environments

A platform for updating one or more properties of one or more digital twins including receiving a request for one or more digital twins; retrieving the one or more digital twins required to fulfill the request from a digital twin datastore; retrieving one or more dynamic models corresponding to one or more properties that are depicted in the one or more digital twins indicated by the request; selecting data sources from a set of available data sources based on the one or more inputs of the one or more dynamic models; obtaining data from selected data sources; determining one or more outputs using the retrieved data as one or more inputs to the one or more dynamic models; and updating the one or more properties of the one or more digital twins based on the one or more outputs of the one or more dynamic models.

Duplicate image detection based on image content

The present technology can analyze a collection of images to generate a high-dimension first representation of each image in the collection of images, and then reduce the dimensionality of the high-dimension first representation of each image by identifying significant features from the high-dimension first representation to yield a reduced-dimension second representation for each image in the collection of images. The present technology can then compare the second representation of any image in the collection of images to the second representation of any other image in the collection of images, wherein images having identical second representation are duplicates.

Systems and methods for automatic model generation

Systems and methods for automatically generating models using machine learning techniques.

Method, device, readable medium and electronic device for identifying traffic light signal

The present disclosure relates to a method, device, computer readable media, and electronic devices for identifying a traffic light signal from an image. The method for identifying a traffic light signal from an image includes extracting, based on a deep neural network, multiple layers of first feature maps corresponding to different layers of the deep neural network from the image. The method includes selecting at least two layers of the first feature maps having different scales from the multiple layers of the first feature maps. The method includes inputting the at least two layers of the first feature maps to a convolution layer having a convolution kernel matching a shape of a traffic light to obtain a second feature map. The method includes obtaining a detection result of the traffic light signal based on the second feature map.

Generic mapping for tracking target object in video sequence

A method of tracking a position of a target object in a video sequence includes identifying the target object in a reference frame. A generic mapping is applied to the target object being tracked. The generic mapping is generated by learning possible appearance variations of a generic object. The method also includes tracking the position of the target object in subsequent frames of the video sequence by determining whether an output of the generic mapping of the target object matches an output of the generic mapping of a candidate object.

Systems and methods for processing content using convolutional neural networks

Systems, methods, and non-transitory computer-readable media can obtain a set of video frames at a first resolution. Process the set of video frames using a convolutional neural network to output one or more signals, the convolutional neural network including (i) a set of two-dimensional convolutional layers and (ii) a set of three-dimensional convolutional layers, wherein the processing causes the set of video frames to be reduced to a second resolution. Process the one or more signals using a set of three-dimensional de-convolutional layers of the convolutional neural network. Obtain one or more outputs corresponding to the set of video frames from the convolutional neural network.

Methods for analyzing and reducing inter/intra site variability using reduced reference images and improving radiologist diagnostic accuracy and consistency

A method of securely accessing an image review unit, including: a triage unit configured to determine if an image of interest is normal or abnormal based upon a reference image and extract normal features from the image of interest based on normal features indicated in the reference image, wherein the reference image and the image of interest are acquired by a same medical imaging device or same doctor or same medical facility; and an image transformation unit configured to reconstruct the image of interest based upon the reference image so as to align the normal features in the image of interest with the normal features in the reference image.

BINOCULAR IMAGE-BASED MODEL TRAINING METHOD AND APPARATUS, AND DATA PROCESSING DEVICE

A binocular image-based model training method and apparatus, and a data processing device are provided. An image matching model includes a teacher model and a student model. In the method, two groups of sample images acquired at different time points by a binocular image acquisition apparatus are first obtained; then, for any two sample images in the two groups of sample images, optical flow estimation is performed according to a preset geometric constraint between the two sample images by means of the teacher model, so as to obtain a more accurate high-confidence optical flow estimation result, the preset geometric constraint being a binocular image-based geometric constraint; and finally, machine learning training of image element matching is performed on the student model by using the two sample image, with the high-confidence optical flow estimation result taken as labeling information. 1. A binocular image-based model training method , applicable to training of an image matching model , with the image matching model comprising a teacher model and a student model , wherein the method comprises steps of:obtaining two groups of sample images acquired by a binocular image acquisition apparatus at different time points;performing, through the teacher model, optical flow estimation, directed at any two sample images in the two groups of sample images, according to a preset geometric constraint between the two sample images, so as to obtain an optical flow estimation result, wherein the preset geometric constraint is a geometric constraint based on binocular images;performing, with the optical flow estimation result as labeling information, machine learning training of image element matching on the student model by using the two sample images, wherein a process of the image element matching is of identifying image elements belonging to a same object in the two sample images.2. The method according to claim 1 , further comprising steps of:obtaining two images to be ...

TARGET DETECTION METHOD BASED ON FUSION OF VISION, LIDAR, AND MILLIMETER WAVE RADAR

A target detection method based on fusion of vision, lidar and millimeter wave radar comprises: obtaining original data detected by a camera, a millimeter wave radar, and a lidar, and synchronizing the millimeter wave radar, the lidar, and the camera in time and space; performing a calculation on the original data detected by the millimeter wave radar according to a radar protocol; generating a region of interest by using a position, a speed, and a radar reflection area obtained from the calculation; extracting feature maps of a point cloud bird's-eye view and the original data detected by the camera; projecting the region of interest onto the feature maps of the point cloud bird's-eye view and the original data detected by the camera; fusing the feature maps of the point cloud bird's-eye view and the original data detected by the camera, and processing a fused image through a fully connected layer. 1. A target detection method based on fusion of vision , lidar , and millimeter wave radar , comprising:(1) obtaining original data detected by each of a camera, a millimeter wave radar, and a lidar, and synchronizing the millimeter wave radar, the lidar, and the camera in time and space;(2) performing a calculation on the original data detected by the millimeter wave radar according to a radar protocol;(3) using a position based on the original data, which is detected by the millimeter wave radar and has been calculated, as a first anchor point, and generating a first region of interest, which is three-dimensional, according to a speed and a radar reflection area with the first anchor point as a center of the first region of interest;(4) generating a second anchor point arranged according to a specified distance in a blind area in which radar points of the millimeter wave radar are not generated, and generating a second region of interest by traversing the second anchor point with the second anchor point as a center of the second region of interest;(5) pre-processing ...

HAND POSE ESTIMATION METHOD, DEVICE AND STORAGE MEDIUM

Provided are a hand pose estimation method, a device and a computer storage medium. The method may include: determining a classification logic map corresponding to each of a plurality of key-points, the plurality of key-points may represent key nodes of a skeleton of a target hand skeleton, a first key-point may be any one of the plurality of key-points; determining, based on a preset classification map and the classification logic map corresponding to the first key-point, co-ordinate information of the first key-point; and obtaining a pose estimation result of the target hand, in response to determining the co-ordinate information corresponding to each of the plurality of key-points. 1. A hand pose estimation method , comprising:determining a classification logic map corresponding to each of a plurality of key-points, wherein the plurality of key-points represent key nodes of a skeleton of a target hand, a first key-point is any one of the plurality of key-points;determining, based on a preset classification map and the classification logic map corresponding to the first key-point, co-ordinate information of the first key-point; andobtaining a pose estimation result of the target hand, after determining co-ordinate information corresponding to each of the plurality of key-points.2. The method as claimed in claim 1 , wherein the determining the classification logic map corresponding to each of the plurality of key-points comprises:acquiring a feature map corresponding to the target hand; andperforming a classification process on the plurality of key-points in the feature map, and obtaining a classification logic map corresponding to each of the plurality of key-points.3. The method as claimed in claim 2 , wherein the acquiring the feature map corresponding to the target hand comprises:acquiring a depth image comprising the target hand;performing a hand detection process on the depth image by a preset feature extractor, and obtaining an initial feature map comprising ...

Attention-based image generation neural networks

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating an output image. In one aspect, one of the methods includes generating the output image intensity value by intensity value according to a generation order of pixel-color channel pairs from the output image, comprising, for each particular generation order position in the generation order: generating a current output image representation of a current output image, processing the current output image representation using a decoder neural network to generate a probability distribution over possible intensity values for the pixel-color channel pair at the particular generation order position, wherein the decoder neural network includes one or more local masked self-attention sub-layers; and selecting an intensity value for the pixel-color channel pair at the particular generation order position using the probability distribution.

System and method for tracking and recognizing people

A tracking and recognition system is provided. The system includes a computer vision-based identity recognition system configured to recognize one or more persons, without a priori knowledge of the respective persons, via an online discriminative learning of appearance signature models of the respective persons. The computer vision-based identity recognition system includes a memory physically encoding one or more routines, which when executed, cause the performance of constructing pairwise constraints between the unlabeled tracking samples. The computer vision-based identity recognition system also includes a processor configured to receive unlabeled tracking samples collected from one or more person trackers and to execute the routines stored in the memory via one or more algorithms to construct the pairwise constraints between the unlabeled tracking samples.

SYSTEM AND METHOD FOR IDENTIFYING LENGTHS OF PARTICLES

A method and system for identifying lengths of a particle in a flow of harvested material comprising particles in an agricultural harvester is disclosed. The agricultural harvester has at least one work assembly for harvesting a crop or for processing harvested material of the crop and can be adjusted using machine parameters. The harvested material is transported as a flow of harvested material through the agricultural harvester while the agricultural harvester is operating. The agricultural harvester has a camera that takes images of the flow of harvested material, with a computing unit of the agricultural harvester analyzing the images of the flow of harvested material in an analytical routine thereby finding particle lengths of particles of the flow of harvested material that are an excess length. The analytical routine is based on a machine learning method trained to find particles with the excess length, with the computing unit using the analytical routine to ascertain an amount of particles with the excess length. 1. A method for identifying lengths of one or more particles in a flow of harvested material comprising the one or more particles , the method comprising:using an agricultural harvester having at least one work assembly for performing one or both of harvesting a crop or processing harvested material of the crop, wherein the at least one work assembly is adjusted using one or more machine parameters;transporting the harvested material as a flow of the harvested material through the agricultural harvester while the agricultural harvester is operating;obtaining, using at least one camera on the agricultural harvester, one or more images of the flow of the harvested material;analyzing, in an analytical routine using a computing unit, the one or more images of the flow of the harvested material to derive particle lengths of the one or more particles of the flow of the harvested material contained in the one or more images,wherein the particle lengths ...

Distributed similarity learning for high-dimensional image features

A system and method for distributed similarity learning for high-dimensional image features are described. A set of data features is accessed. Subspaces from a space formed by the set of data features are determined using a set of projection matrices. Each subspace has a dimension lower than a dimension of the set of data features. Similarity functions are computed for the subspaces. Each similarity function is based on the dimension of the corresponding subspace. A linear combination of the similarity functions is performed to determine a similarity function for the set of data features.

Method for grouping cells according to density and electronic device employing method

A method of grouping certain cell densities to establish the number and volume of cells appearing in an image input the image into a self-encoder having a preset number of a density grouping models to obtain a preset number of reconstructed images. The image and each reconstructed image are input into a twin network model of the density grouping model corresponding to each reconstructed image, and a first error value is calculated between the image and each reconstructed image. A minimum first error value in the first error value set is determined, and a density range corresponding to the density grouping model corresponding to minimum first error value is taken as the density range. An electronic device and a non-volatile storage medium performing the above-described method are also disclosed.

Method of chekcing the appearance of the surface of a tyre

Detection method according to Claim 9 , in which, the grey level image of the inner surface of a tyre to be checked is captured and this original image is transformed into an orthonormal space in which the x-axis (OX) represents the circumferential direction, and the y-axis (OY), the predetermined series of filters is applied to the original image of the inner surface of the tyre to be checked in order to obtain a multivariate original image, the multivariate original image is split according to the predefined tiling in the axial and circumferential directions, in such a way as to obtain multivariate sub-images of the inner surface of the tyre to be checked, each of the multivariate sub-images is transformed into one-dimensional vectors, using each of the selected descriptors, in such a way as to obtain a simplified multivariate image of the inner surface of the tyre to be checked, the simplified multivariate image is transformed into the common reduced factorial space, the sub-images of the inner surface of the tyre to be checked containing an anomaly are located using a classifier suitable for identifying the areas of the common reduced factorial space containing the anomalies.

POINT CLOUD DATA PROCESSING METHOD AND DEVICE, COMPUTER DEVICE, AND STORAGE MEDIUM

A point cloud data processing method and device, a computer device and a storage medium are provided. The method includes: acquiring point cloud data, and constructing a corresponding neighboring point set for each of data points in the point cloud data; calculating Hausdorff distances between the neighboring point set and a pre-constructed kernel point cloud to obtain a distance matrix; calculating a convolution of the neighboring point set with the distance matrix and a network weight matrix in a Hausdorff convolution layer in an encoder, to obtain high-dimensional point cloud features, the encoder and a decoder being two parts in a deep learning network; and reducing feature dimension of the high-dimensional point cloud features through the decoder, so that a classifier performs semantic classification on the point cloud data according to object point cloud features obtained by the dimension reduction. 1. A point cloud data processing method , comprising:acquiring point cloud data, and constructing a corresponding neighboring point set for each of data points in the point cloud data;calculating Hausdorff distances between the neighboring point set and a pre-constructed kernel point cloud to obtain a distance matrix;calculating a convolution of the neighboring point set with the distance matrix and a network weight matrix in a Hausdorff convolution layer in an encoder to obtain high-dimensional point cloud features; the encoder and a decoder being two parts in a deep learning network; andreducing feature dimension of the high-dimensional point cloud features through the decoder, so that a classifier performs semantic classification on the point cloud data according to object point cloud features obtained by the dimension reduction.2. The point cloud data processing method according to claim 1 , wherein the number of Hausdorff convolutional layers in the encoder is not less than two; and the calculating a convolution of the neighboring point set with the distance ...

MINIATURE MICROSCOPIC CELL IMAGE ACQUISITION DEVICE AND IMAGE RECOGNITION METHOD

A miniature microscopic cell image acquisition device and an image recognition method are provided. The miniature microscopic cell image acquisition device comprises a support, wherein a movable module platform is provided on the support, and a camera module is provided on the module platform. A microscope head that is relatively fixed is provided below a camera of the camera module, a slide holder is provided below the microscope head, and a lighting source is provided below the slide holder. A scanning drive module is provided between the slide holder and the camera module to perform a scanning movement along X and Y axes, so that the slide holder and the camera module make a scanning movement along the X and Y axes, and images of a slide are acquired by the camera module in a scanning manner. 14. A miniature microscopic cell image acquisition device , comprising a support () , wherein{'b': 2', '4', '1', '2, 'a movable module platform () is provided on the support (), and a camera module () is provided on the module platform ();'}{'b': 3', '111', '1', '5', '3', '8', '5, 'a microscope head () that is relatively fixed is provided below a camera () of the camera module (), a slide holder () is provided below the microscope head (), and a lighting source () is provided below the slide holder (); and'}{'b': 5', '1', '5', '1', '7', '1, 'a scanning drive module is provided between the slide holder () and the camera module () to perform a scanning movement along X axis and Y axis, so that the slide holder () and the camera module () make a scanning movement along the X axis and Y axis, and images of a slide () are collected by the camera module () in a scanning manner.'}2. The miniature microscopic cell image acquisition device according to claim 1 , wherein claim 1 ,{'b': 3', '32', '2', '32', '2, 'the microscope head () comprises a cantilever rod () mounted on the module platform (), one end of the cantilever rod () is fixedly connected to the module platform (), and a ...

APPARATUS AND METHOD FOR DETECTING FACIAL POSE, IMAGE PROCESSING SYSTEM, AND STORAGE MEDIUM

The present disclosure discloses an apparatus and a method for detecting a facial pose, an image processing system, and a storage medium. The apparatus comprises: an obtaining unit to obtain at least three keypoints of at least one face from an input image based on a pre-generated neural network, wherein coordinates of the keypoints obtained via a layer in the neural network for obtaining coordinates are three-dimensional coordinates; and a determining unit to determine, for the at least one face, a pose of the face based on the obtained keypoints, wherein the determined facial pose includes at least an angle. According to the present disclosure, the accuracy of the three-dimensional coordinates of the facial keypoints can be improved, thus the detection precision of a facial pose can be improved. 1. An apparatus for detecting a facial pose , comprising:an obtaining unit configured to obtain at least three keypoints of at least one face from an input image based on a neural network, wherein coordinates of the keypoints obtained via a layer in the neural network for obtaining coordinates are three-dimensional coordinates; anda determining unit configured to determine, for the at least one face, a pose of the face based on the obtained keypoints, wherein the determined facial pose includes at least an angle in one dimension.2. The apparatus according to claim 1 , wherein in a case where the input image includes faces of different scales claim 1 , the obtaining unit obtains at least three keypoints of each face simultaneously based on the neural network.3. The apparatus according to claim 2 , wherein the neural network is pre-generated by:acquiring a sample image in which faces are labeled with keypoints and acquiring predefined reference regions that can cover the faces in the sample image; wherein the sample image includes at least one face, wherein for a face, at least three keypoints are labeled, and coordinates of the labeled keypoints are three-dimensional ...

SYSTEMS AND METHODS FOR PROCESSING ELECTRONIC IMAGES TO DETERMINE TESTING FOR UNSTAINED SPECIMENS

A computer-implemented method may include receiving a collection of unstained digital histopathology slide images at a storage device and running a trained machine learning model on one or more slide images of the collection to infer a presence or an absence of a salient feature. The trained machine learning model may have been trained by processing a second collection of unstained or stained digital histopathology slide images and at least one synoptic annotation for one or more unstained or stained digital histopathology slide images of the second collection. The computer-implemented method may further include determining at least one map from output of the trained machine learning model and providing an output from the trained machine learning model to the storage device. 1. A computer-implemented method , comprising:receiving a collection of unstained digital histopathology slide images at a storage device; 'wherein the trained machine learning model has been trained by processing a second collection of unstained or stained digital histopathology slide images and at least one synoptic annotation for one or more unstained or stained digital histopathology slide images of the second collection;', 'running a trained machine learning model on one or more slide images of the collection to infer a presence or an absence of a salient feature,'} determining the at least one map using an introspection technique if the output comprises a presence or absence of the salient feature,', 'determining the at least one map using the output if the output comprises instance/semantic segmentations, and', 'determining the at least one map by indicating a location of the salient feature if the output comprises detection regions; and, 'determining at least one map based on output of the trained machine learning model, the determining of the at least one map further comprisingproviding an output from the trained machine learning model to the storage device.2. The computer-implemented ...

Image classifying method and image displaying method

An image classifying method includes the following operations: reducing an order of a color of a first image to generate a first order reduction image, wherein the first order reduction image includes several first image blocks; obtaining a second order reduction image from a database, wherein the second order reduction image includes several second image blocks; calculating several block color difference values between the first order reduction image and the second order reduction image respectively according to differences between a color value of each of the first image blocks and a color value of each of the second image blocks; and determining whether or not the first image belongs to a same category as the second order reduction image according to the block color difference values between the first order reduction image and the second order reduction image.

Method, System, and Computer Program Product for Data Pre-Processing in Deep Learning

The goal of this invention is to develop smart and fast data processing scheme for more computational efficient deep learning to support adaptive and real-time applications. We propose to apply Singular-Value Decomposition (SVD)-QR algorithm to preprocessing of deep learning for large scale data input. For the mass data input, we apply Limited Memory Subspace Optimization for SVD (LMSVD)-QR algorithm to increase the data processing speed. Simulation results in automated handwritten digit recognition show that SVD-QR and LMSVD-QR can tremendously reduce the number of input to deep learning neural network without losing its performance, and both can tremendously increase the data processing speed for deep learning.

Identifying scene boundaries using group sparsity analysis

A method for identifying a set of key video frames from a video sequence comprising extracting feature vectors for each video frame and applying a group sparsity algorithm to represent the feature vector for a particular video frame as a group sparse combination of the feature vectors for the other video frames. Weighting coefficients associated with the group sparse combination are analyzed to determine video frame clusters of temporally-contiguous, similar video frames. The video sequence is segmented into scenes by identifying scene boundaries based on the determined video frame clusters.

STRESS COPING STYLE DETERMINATION SYSTEM, STRESS COPING STYLE DETERMINATION METHOD, LEARNING DEVICE, LEARNING METHOD, PROGRAM, AND LEARNED MODEL

Provided is a system capable of determining a stress coping style of a test subject in a non-contact state. The system includes a biological information acquiring part which acquires biological information of a test subject in a non-contact state, and a determining part which determines a stress coping style of the test subject based on the biological information and a response pattern specified in advance. The response pattern is specified by a hemodynamic parameter. 1. A stress coping style determination system , comprising:a biological information acquiring part which acquires biological information of a test subject in a non-contact state; anda determining part which determines a stress coping style of the test subject based on the biological information and a response pattern specified in advance, whereinthe response pattern is specified by a hemodynamic parameter.2. The stress coping style determination system according to claim 1 , whereinthe hemodynamic parameter includes a plurality of parameters among a mean blood pressure, a heart rate, a cardiac output, a stroke volume, and a total peripheral resistance.3. The stress coping style determination system according to claim 1 , whereinthe biological information is a facial image.4. The stress coping style determination system according to claim 3 , whereinthe facial image is a facial thermal image or a facial visible image.5. The stress coping style determination system according to claim 1 , whereinthe determining part determines the stress coping style of the test subject by observing a stress response of a specific region of a facial surface including in the facial image.6. The stress coping style determination system according to claim 5 , whereinthe response pattern includes patterns of three types: “active coping”, “passive coping”, and “no coping”.7. The stress coping style determination system according to claim 6 , whereinthe determining part comprises a determination-purpose feature value storage ...

CONTROL METHOD AND INFORMATION PROCESSING APPARATUS

A storage unit holds a classification model that calculates a confidence score from image data, and a transformation model that is a model for transforming a feature value having fewer dimensions than the image data into the image data and is created such that a set of feature values corresponding to a set of image data follows a probability distribution. A processing unit extracts a feature value according to the probability distribution. The processing unit transforms the feature value into image data using the transformation model and calculates a confidence score corresponding to the image data using the classification model. The processing unit updates, based on the probability distribution and the feature value, a feature value to be input to the transformation model from the feature value to a feature value in such a manner that a confidence score to be calculated is higher than the confidence score. 1. A control method comprising:obtaining, by a processor, a classification model and a transformation model, the classification model being configured to calculate, from input image data, a confidence score indicating a likelihood that the input image data belongs to a specified class, the transformation model being a model for transforming an input feature value having fewer dimensions than the input image data into the input image data and being created such that a set of feature values corresponding to a set of image data follows a specific probability distribution;extracting, by the processor, a first feature value according to the specific probability distribution;transforming, by the processor, the first feature value into first image data using the transformation model, and calculating a first confidence score corresponding to the first image data using the classification model; andupdating, by the processor, based on the specific probability distribution and the first feature value, a feature value to be input to the transformation model from the first ...

TARGET OBJECT DETECTION DEVICE, TARGET OBJECT DETECTION METHOD, AND NON-TRANSITORY COMPUTER READABLE STORAGE MEDIUM STORING TARGET OBJECT DETECTION PROGRAM

A target object detection device comprising a camera unit that acquires an image at a predetermined time interval; an image processing unit that extracts a target object from the image acquired; a comparison unit that compares the target object extracted from the image with a target object extracted from an image of a frame before the image; a storage unit that stores a weighting condition based on a position on an image between the target object extracted from the image and the target object extracted from the image of a frame before the image; and an identification unit that compares a comparison result of the comparison unit based on the weighting condition and identifies the target object extracted from the image of a frame before the image that matches the target object extracted from the image. 1. A target object detection device comprising:a camera unit that acquires an image at a predetermined time interval;an image processing unit that extracts a target object from the image acquired;a comparison unit that compares the target object extracted from the image with a target object extracted from an image of a frame before the image;a storage unit that stores a weighting condition based on a position on an image between the target object extracted from the image and the target object extracted from the image of a frame before the image; andan identification unit that compares a comparison result of the comparison unit based on the weighting condition and identifies the target object extracted from the image of a frame before the image that matches the target object extracted from the image.2. The target object detection device according to claim 1 , whereinthe comparison unit calculates a relationship of target objects to be compared based on a comparison result, andthe identification unit identifies target objects having a high relationship as an identical target object.3. The target object detection device according to claim 1 , whereinthe weighting condition ...

RECURRENT MULTI-TASK CONVOLUTIONAL NEURAL NETWORK ARCHITECTURE

A recurrent multi-task CNN with an encoder and multiple decoders infers single value output and dense (image) outputs such as heatmaps and segmentation masks. Recurrence is obtained by reinjecting (with mere concatenation) heatmaps or masks (or intermediate feature maps) to a next input image (or to next intermediate feature maps) for a next CNN inference. The inference outputs may be refined using cascaded refiner blocks specifically trained. Virtual annotation for training video sequences can be obtained using computer analysis. Benefits of these approaches allows the depth of the CNN, i.e. the number of layers, to be reduced. They also avoid parallel independent inferences to be run for different tasks, while keeping similar prediction quality. Multiple task inferences are useful for Augmented Reality applications. 1. An image processing device comprising a processor-implemented neural network , the neural network comprising:an input block configured to obtain successive inference inputs from successive input images forming an input sequence,a layer-based neural encoder configured to determine, during an inference of the processor-implemented neural network, feature maps from one of the inference inputs, andmultiple layer-based neural decoders, each having at least one separate layer not shared with the other layer-based neural decoder or decoders, configured to generate, during the inference, multiple respective inference outputs from the feature maps, wherein at least one of the inference outputs is an inference image of image type spatially characterizing image content of the input image,wherein the inference input of the layer-based neural encoder for a next inference of the processor-implemented neural network is built from a next input image of the input sequence and from at least one image-type inference output generated during a previous inference based on a previous image in the input sequence.2. The image processing device of claim 1 , wherein the image ...

Multi-task self-training for learning general representations

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training an image representation neural network.

METHOD AND SYSTEM FOR SEMI-SUPERVISED CONTENT LOCALIZATION

A special-purpose convolutional learning model architecture outputs a convolutional feature map at a last of its convolutional layers, then performs binary classification based on non-semantically labeled dataset. The convolutional feature map, containing a combination of low-spatial resolution features and high-spatial resolution features, in conjunction with a binary classification output of a special-purpose learning model having transferred learning from a pre-trained learning model, may be used to non-semantically derive a segmentation map. The segmentation map may reflect both low-spatial resolution and high-spatial resolution features of the original image on a one-to-one pixel correspondence, and thus may be utilized to highlight or obscure subject matter of the image in a contextually fitting manner at both a global scale and a local scale over the image, without semantic knowledge of the content of the image. 1. A method comprising:deriving a gradient of a feature vector of a convolutional feature map;deriving a feature map contribution parameter with regard to the feature vector from the gradient of the feature vector;obtaining a localization map by aggregating feature map contribution parameters; anddrawing an edge of a segmentation map based on values of the localization map.2. The method of claim 1 , wherein the gradient of the feature vector is derived with regard to a classification.3. The method of claim 2 , wherein deriving the gradient of the feature vector comprises computing a partial derivative of a probability score of a feature of the convolutional feature map with regard to the classification over a feature vector of the convolutional feature map.4. The method of claim 3 , wherein deriving the feature map contribution parameter comprises normalizing the partial derivative over each pixel of the convolutional feature map.5. The method of claim 4 , wherein normalizing the partial derivative comprises summing the partial derivative over each ...

Identity Authentication Method, Terminal Device And System

An identity authentication method, terminal device and system are provided. A terminal device scans a two-dimensional code after receiving a two-dimensional code scanning instruction. After receiving the two-dimensional code scanning instruction, the terminal device automatically opens a biometric information acquisition function, and acquires the biometric information of a user currently operating the terminal device. Identity authentication is performed on the biometric information, to determine whether the user has an operation authority corresponding to the two-dimensional code.

Method, system and apparatus for associating a target object in images

Image association methods, systems and apparatuses are provided. The method includes: obtaining a first image and a second image, where the first image is obtained by capturing a scene by a first image capture device at a first view, the second image is obtained by synchronously capturing the scene by a second image capture device at a second view, and the first view is different from the second view; determining an epipole of the first image capture device on a plane of the second image; determining a projection point of a first target point in a first bounding box on the second image, where the first bounding box is a bounding box of a target in the first image; and determining an association bounding box of the first bounding box in the second image according to the epipole and the projection point.